2                          Debugging on Linux for s/390 & z/Architecture
   3                                       by
   4                Denis Joseph Barrow (,
   5                Copyright (C) 2000-2001 IBM Deutschland Entwicklung GmbH, IBM Corporation
   6                              Best viewed with fixed width fonts 
   8Overview of Document:
  10This document is intended to give a good overview of how to debug
  11Linux for s/390 & z/Architecture. It isn't intended as a complete reference & not a
  12tutorial on the fundamentals of C & assembly. It doesn't go into
  13390 IO in any detail. It is intended to complement the documents in the
  14reference section below & any other worthwhile references you get.
  16It is intended like the Enterprise Systems Architecture/390 Reference Summary
  17to be printed out & used as a quick cheat sheet self help style reference when
  18problems occur.
  22Register Set
  23Address Spaces on Intel Linux
  24Address Spaces on Linux for s/390 & z/Architecture
  25The Linux for s/390 & z/Architecture Kernel Task Structure
  26Register Usage & Stackframes on Linux for s/390 & z/Architecture
  27A sample program with comments
  28Compiling programs for debugging on Linux for s/390 & z/Architecture
  29Figuring out gcc compile errors
  30Debugging Tools
  33Performance Debugging 
  34Debugging under VM
  35s/390 & z/Architecture IO Overview
  36Debugging IO on s/390 & z/Architecture under VM
  37GDB on s/390 & z/Architecture
  38Stack chaining in gdb by hand
  39Examining core dumps
  41Debugging modules
  42The proc file system
  43Starting points for debugging scripting languages etc.
  46Special Thanks
  48Register Set
  50The current architectures have the following registers.
  5216  General propose registers, 32 bit on s/390 64 bit on z/Architecture, r0-r15 or gpr0-gpr15 used for arithmetic & addressing. 
  5416 Control registers, 32 bit on s/390 64 bit on z/Architecture, ( cr0-cr15 kernel usage only ) used for memory management,
  55interrupt control,debugging control etc.
  5716 Access registers ( ar0-ar15 ) 32 bit on s/390 & z/Architecture
  58not used by normal programs but potentially could 
  59be used as temporary storage. Their main purpose is their 1 to 1
  60association with general purpose registers and are used in
  61the kernel for copying data between kernel & user address spaces.
  62Access register 0 ( & access register 1 on z/Architecture ( needs 64 bit 
  63pointer ) ) is currently used by the pthread library as a pointer to
  64the current running threads private area.
  6616 64 bit floating point registers (fp0-fp15 ) IEEE & HFP floating 
  67point format compliant on G5 upwards & a Floating point control reg (FPC) 
  684  64 bit registers (fp0,fp2,fp4 & fp6) HFP only on older machines.
  70Linux (currently) always uses IEEE & emulates G5 IEEE format on older machines,
  71( provided the kernel is configured for this ).
  74The PSW is the most important register on the machine it
  75is 64 bit on s/390 & 128 bit on z/Architecture & serves the roles of 
  76a program counter (pc), condition code register,memory space designator.
  77In IBM standard notation I am counting bit 0 as the MSB.
  78It has several advantages over a normal program counter
  79in that you can change address translation & program counter 
  80in a single instruction. To change address translation,
  81e.g. switching address translation off requires that you
  82have a logical=physical mapping for the address you are
  83currently running at.
  85      Bit           Value
  86s/390 z/Architecture
  870       0     Reserved ( must be 0 ) otherwise specification exception occurs.
  891       1     Program Event Recording 1 PER enabled, 
  90              PER is used to facilitate debugging e.g. single stepping.
  922-4    2-4    Reserved ( must be 0 ). 
  945       5     Dynamic address translation 1=DAT on.
  966       6     Input/Output interrupt Mask
  987       7     External interrupt Mask used primarily for interprocessor signalling & 
  99              clock interrupts.
 1018-11  8-11    PSW Key used for complex memory protection mechanism not used under linux
 10312      12    1 on s/390 0 on z/Architecture
 10513      13    Machine Check Mask 1=enable machine check interrupts
 10714      14    Wait State set this to 1 to stop the processor except for interrupts & give 
 108              time to other LPARS used in CPU idle in the kernel to increase overall 
 109              usage of processor resources.
 11115      15    Problem state ( if set to 1 certain instructions are disabled )
 112              all linux user programs run with this bit 1 
 113              ( useful info for debugging under VM ).
 11516-17 16-17   Address Space Control
 117              00 Primary Space Mode when DAT on
 118              The linux kernel currently runs in this mode, CR1 is affiliated with 
 119              this mode & points to the primary segment table origin etc.
 121              01 Access register mode this mode is used in functions to 
 122              copy data between kernel & user space.
 124              10 Secondary space mode not used in linux however CR7 the
 125              register affiliated with this mode is & this & normally
 126              CR13=CR7 to allow us to copy data between kernel & user space.
 127              We do this as follows:
 128              We set ar2 to 0 to designate its
 129              affiliated gpr ( gpr2 )to point to primary=kernel space.
 130              We set ar4 to 1 to designate its
 131              affiliated gpr ( gpr4 ) to point to secondary=home=user space
 132              & then essentially do a memcopy(gpr2,gpr4,size) to
 133              copy data between the address spaces, the reason we use home space for the
 134              kernel & don't keep secondary space free is that code will not run in 
 135              secondary space.
 137              11 Home Space Mode all user programs run in this mode.
 138              it is affiliated with CR13.
 14018-19 18-19   Condition codes (CC)
 14220    20      Fixed point overflow mask if 1=FPU exceptions for this event 
 143              occur ( normally 0 ) 
 14521    21      Decimal overflow mask if 1=FPU exceptions for this event occur 
 146              ( normally 0 )
 14822    22      Exponent underflow mask if 1=FPU exceptions for this event occur 
 149              ( normally 0 )
 15123    23      Significance Mask if 1=FPU exceptions for this event occur 
 152              ( normally 0 )
 15424-31 24-30   Reserved Must be 0.
 156      31      Extended Addressing Mode
 157      32      Basic Addressing Mode
 158              Used to set addressing mode
 159              PSW 31   PSW 32
 160                0         0        24 bit
 161                0         1        31 bit
 162                1         1        64 bit
 16432             1=31 bit addressing mode 0=24 bit addressing mode (for backward 
 165               compatibility), linux always runs with this bit set to 1
 16733-64          Instruction address.
 168      33-63    Reserved must be 0
 169      64-127   Address
 170               In 24 bits mode bits 64-103=0 bits 104-127 Address 
 171               In 31 bits mode bits 64-96=0 bits 97-127 Address
 172               Note: unlike 31 bit mode on s/390 bit 96 must be zero
 173               when loading the address with LPSWE otherwise a 
 174               specification exception occurs, LPSW is fully backward
 175               compatible.
 178Prefix Page(s)
 180This per cpu memory area is too intimately tied to the processor not to mention.
 181It exists between the real addresses 0-4096 on s/390 & 0-8192 z/Architecture & is exchanged 
 182with a 1 page on s/390 or 2 pages on z/Architecture in absolute storage by the set 
 183prefix instruction in linux'es startup. 
 184This page is mapped to a different prefix for each processor in an SMP configuration
 185( assuming the os designer is sane of course :-) ).
 186Bytes 0-512 ( 200 hex ) on s/390 & 0-512,4096-4544,4604-5119 currently on z/Architecture 
 187are used by the processor itself for holding such information as exception indications & 
 188entry points for exceptions.
 189Bytes after 0xc00 hex are used by linux for per processor globals on s/390 & z/Architecture 
 190( there is a gap on z/Architecture too currently between 0xc00 & 1000 which linux uses ).
 191The closest thing to this on traditional architectures is the interrupt
 192vector table. This is a good thing & does simplify some of the kernel coding
 193however it means that we now cannot catch stray NULL pointers in the
 194kernel without hard coded checks.
 198Address Spaces on Intel Linux
 201The traditional Intel Linux is approximately mapped as follows forgive
 202the ascii art.
 2030xFFFFFFFF 4GB Himem                        *****************
 204                                            *               *
 205                                            * Kernel Space  *
 206                                            *               *
 207                                            *****************          ****************
 208User Space Himem (typically 0xC0000000 3GB )*  User Stack   *          *              *
 209                                            *****************          *              *
 210                                            *  Shared Libs  *          * Next Process *          
 211                                            *****************          *     to       *  
 212                                            *               *    <==   *     Run      *  <==
 213                                            *  User Program *          *              *
 214                                            *   Data BSS    *          *              *
 215                                            *    Text       *          *              *
 216                                            *   Sections    *          *              *
 2170x00000000                                  *****************          ****************
 219Now it is easy to see that on Intel it is quite easy to recognise a kernel address 
 220as being one greater than user space himem ( in this case 0xC0000000).
 221& addresses of less than this are the ones in the current running program on this
 222processor ( if an smp box ).
 223If using the virtual machine ( VM ) as a debugger it is quite difficult to
 224know which user process is running as the address space you are looking at
 225could be from any process in the run queue.
 227The limitation of Intels addressing technique is that the linux
 228kernel uses a very simple real address to virtual addressing technique
 229of Real Address=Virtual Address-User Space Himem.
 230This means that on Intel the kernel linux can typically only address
 231Himem=0xFFFFFFFF-0xC0000000=1GB & this is all the RAM these machines
 232can typically use.
 233They can lower User Himem to 2GB or lower & thus be
 234able to use 2GB of RAM however this shrinks the maximum size
 235of User Space from 3GB to 2GB they have a no win limit of 4GB unless
 236they go to 64 Bit.
 239On 390 our limitations & strengths make us slightly different.
 240For backward compatibility we are only allowed use 31 bits (2GB)
 241of our 32 bit addresses, however, we use entirely separate address 
 242spaces for the user & kernel.
 244This means we can support 2GB of non Extended RAM on s/390, & more
 245with the Extended memory management swap device & 
 246currently 4TB of physical memory currently on z/Architecture.
 249Address Spaces on Linux for s/390 & z/Architecture
 252Our addressing scheme is as follows
 255Himem 0x7fffffff 2GB on s/390    *****************          ****************
 256currently 0x3ffffffffff (2^42)-1 *  User Stack   *          *              *
 257on z/Architecture.               *****************          *              *
 258                                 *  Shared Libs  *          *              *      
 259                                 *****************          *              *  
 260                                 *               *          *    Kernel    *  
 261                                 *  User Program *          *              *
 262                                 *   Data BSS    *          *              *
 263                                 *    Text       *          *              *
 264                                 *   Sections    *          *              *
 2650x00000000                       *****************          ****************
 267This also means that we need to look at the PSW problem state bit
 268or the addressing mode to decide whether we are looking at
 269user or kernel space.
 271Virtual Addresses on s/390 & z/Architecture
 274A virtual address on s/390 is made up of 3 parts
 275The SX ( segment index, roughly corresponding to the PGD & PMD in linux terminology ) 
 276being bits 1-11.
 277The PX ( page index, corresponding to the page table entry (pte) in linux terminology )
 278being bits 12-19. 
 279The remaining bits BX (the byte index are the offset in the page )
 280i.e. bits 20 to 31.
 282On z/Architecture in linux we currently make up an address from 4 parts.
 283The region index bits (RX) 0-32 we currently use bits 22-32
 284The segment index (SX) being bits 33-43
 285The page index (PX) being bits  44-51
 286The byte index (BX) being bits  52-63
 2891) s/390 has no PMD so the PMD is really the PGD also.
 290A lot of this stuff is defined in pgtable.h.
 2922) Also seeing as s/390's page indexes are only 1k  in size 
 293(bits 12-19 x 4 bytes per pte ) we use 1 ( page 4k )
 294to make the best use of memory by updating 4 segment indices 
 295entries each time we mess with a PMD & use offsets 
 2960,1024,2048 & 3072 in this page as for our segment indexes.
 297On z/Architecture our page indexes are now 2k in size
 298( bits 12-19 x 8 bytes per pte ) we do a similar trick
 299but only mess with 2 segment indices each time we mess with
 300a PMD.
 3023) As z/Architecture supports up to a massive 5-level page table lookup we
 303can only use 3 currently on Linux ( as this is all the generic kernel
 304currently supports ) however this may change in future
 305this allows us to access ( according to my sums )
 3064TB of virtual storage per process i.e.
 3074096*512(PTES)*1024(PMDS)*2048(PGD) = 4398046511104 bytes,
 308enough for another 2 or 3 of years I think :-).
 309to do this we use a region-third-table designation type in
 310our address space control registers.
 313The Linux for s/390 & z/Architecture Kernel Task Structure
 315Each process/thread under Linux for S390 has its own kernel task_struct
 316defined in linux/include/linux/sched.h
 317The S390 on initialisation & resuming of a process on a cpu sets
 318the __LC_KERNEL_STACK variable in the spare prefix area for this cpu
 319(which we use for per-processor globals).
 321The kernel stack pointer is intimately tied with the task structure for
 322each processor as follows.
 324                      s/390
 325            ************************
 326            *  1 page kernel stack *
 327            *        ( 4K )        *
 328            ************************
 329            *   1 page task_struct *        
 330            *        ( 4K )        *
 3318K aligned  ************************ 
 333                 z/Architecture
 334            ************************
 335            *  2 page kernel stack *
 336            *        ( 8K )        *
 337            ************************
 338            *  2 page task_struct  *        
 339            *        ( 8K )        *
 34016K aligned ************************ 
 342What this means is that we don't need to dedicate any register or global variable
 343to point to the current running process & can retrieve it with the following
 344very simple construct for s/390 & one very similar for z/Architecture.
 346static inline struct task_struct * get_current(void)
 348        struct task_struct *current;
 349        __asm__("lhi   %0,-8192\n\t"
 350                "nr    %0,15"
 351                : "=r" (current) );
 352        return current;
 355i.e. just anding the current kernel stack pointer with the mask -8192.
 356Thankfully because Linux doesn't have support for nested IO interrupts
 357& our devices have large buffers can survive interrupts being shut for 
 358short amounts of time we don't need a separate stack for interrupts.
 363Register Usage & Stackframes on Linux for s/390 & z/Architecture
 367This is the code that gcc produces at the top & the bottom of
 368each function. It usually is fairly consistent & similar from 
 369function to function & if you know its layout you can probably
 370make some headway in finding the ultimate cause of a problem
 371after a crash without a source level debugger.
 373Note: To follow stackframes requires a knowledge of C or Pascal &
 374limited knowledge of one assembly language.
 376It should be noted that there are some differences between the
 377s/390 & z/Architecture stack layouts as the z/Architecture stack layout didn't have
 378to maintain compatibility with older linkage formats.
 383This is a built in compiler function for runtime allocation
 384of extra space on the callers stack which is obviously freed
 385up on function exit ( e.g. the caller may choose to allocate nothing
 386of a buffer of 4k if required for temporary purposes ), it generates 
 387very efficient code ( a few cycles  ) when compared to alternatives 
 388like malloc.
 390automatics: These are local variables on the stack,
 391i.e they aren't in registers & they aren't static.
 394This is a pointer to the stack pointer before entering a
 395framed functions ( see frameless function ) prologue got by 
 396dereferencing the address of the current stack pointer,
 397 i.e. got by accessing the 32 bit value at the stack pointers
 398current location.
 401This is a pointer to the back of the literal pool which
 402is an area just behind each procedure used to store constants
 403in each function.
 405call-clobbered: The caller probably needs to save these registers if there 
 406is something of value in them, on the stack or elsewhere before making a 
 407call to another procedure so that it can restore it later.
 410The code generated by the compiler to return to the caller.
 413A frameless function in Linux for s390 & z/Architecture is one which doesn't 
 414need more than the register save area ( 96 bytes on s/390, 160 on z/Architecture )
 415given to it by the caller.
 416A frameless function never:
 4171) Sets up a back chain.
 4182) Calls alloca.
 4193) Calls other normal functions
 4204) Has automatics.
 423This is a pointer to the global-offset-table in ELF
 424( Executable Linkable Format, Linux'es most common executable format ),
 425all globals & shared library objects are found using this pointer.
 428ELF shared libraries are typically only loaded when routines in the shared
 429library are actually first called at runtime. This is lazy binding.
 432This is a table found from the GOT which contains pointers to routines
 433in other shared libraries which can't be called to by easier means.
 436The code generated by the compiler to set up the stack frame.
 439This is extra area allocated on the stack of the calling function if the
 440parameters for the callee's cannot all be put in registers, the same
 441area can be reused by each function the caller calls.
 444A COFF  executable format based concept of a procedure reference 
 445actually being 8 bytes or more as opposed to a simple pointer to the routine.
 446This is typically defined as follows
 447Routine Descriptor offset 0=Pointer to Function
 448Routine Descriptor offset 4=Pointer to Table of Contents
 449The table of contents/TOC is roughly equivalent to a GOT pointer.
 450& it means that shared libraries etc. can be shared between several
 451environments each with their own TOC.
 454static-chain: This is used in nested functions a concept adopted from pascal 
 455by gcc not used in ansi C or C++ ( although quite useful ), basically it
 456is a pointer used to reference local variables of enclosing functions.
 457You might come across this stuff once or twice in your lifetime.
 460The function below should return 11 though gcc may get upset & toss warnings 
 461about unused variables.
 462int FunctionA(int a)
 464        int b;
 465        FunctionC(int c)
 466        {
 467                b=c+1;
 468        }
 469        FunctionC(10);
 470        return(b);
 474s/390 & z/Architecture Register usage
 476r0       used by syscalls/assembly                  call-clobbered
 477r1       used by syscalls/assembly                  call-clobbered
 478r2       argument 0 / return value 0                call-clobbered
 479r3       argument 1 / return value 1 (if long long) call-clobbered
 480r4       argument 2                                 call-clobbered
 481r5       argument 3                                 call-clobbered
 482r6       argument 4                                 saved
 483r7       pointer-to arguments 5 to ...              saved      
 484r8       this & that                                saved
 485r9       this & that                                saved
 486r10      static-chain ( if nested function )        saved
 487r11      frame-pointer ( if function used alloca )  saved
 488r12      got-pointer                                saved
 489r13      base-pointer                               saved
 490r14      return-address                             saved
 491r15      stack-pointer                              saved
 493f0       argument 0 / return value ( float/double ) call-clobbered
 494f2       argument 1                                 call-clobbered
 495f4       z/Architecture argument 2                  saved
 496f6       z/Architecture argument 3                  saved
 497The remaining floating points
 498f1,f3,f5 f7-f15 are call-clobbered.
 5021) The only requirement is that registers which are used
 503by the callee are saved, e.g. the compiler is perfectly
 504capable of using r11 for purposes other than a frame a
 505frame pointer if a frame pointer is not needed.
 5062) In functions with variable arguments e.g. printf the calling procedure 
 507is identical to one without variable arguments & the same number of 
 508parameters. However, the prologue of this function is somewhat more
 509hairy owing to it having to move these parameters to the stack to
 510get va_start, va_arg & va_end to work.
 5113) Access registers are currently unused by gcc but are used in
 512the kernel. Possibilities exist to use them at the moment for
 513temporary storage but it isn't recommended.
 5144) Only 4 of the floating point registers are used for
 515parameter passing as older machines such as G3 only have only 4
 516& it keeps the stack frame compatible with other compilers.
 517However with IEEE floating point emulation under linux on the
 518older machines you are free to use the other 12.
 5195) A long long or double parameter cannot be have the 
 520first 4 bytes in a register & the second four bytes in the 
 521outgoing args area. It must be purely in the outgoing args
 522area if crossing this boundary.
 5236) Floating point parameters are mixed with outgoing args
 524on the outgoing args area in the order the are passed in as parameters.
 5257) Floating point arguments 2 & 3 are saved in the outgoing args area for 
 529Stack Frame Layout
 531s/390     z/Architecture
 5320         0             back chain ( a 0 here signifies end of back chain )
 5334         8             eos ( end of stack, not used on Linux for S390 used in other linkage formats )
 5348         16            glue used in other s/390 linkage formats for saved routine descriptors etc.
 53512        24            glue used in other s/390 linkage formats for saved routine descriptors etc.
 53616        32            scratch area
 53720        40            scratch area
 53824        48            saved r6 of caller function
 53928        56            saved r7 of caller function
 54032        64            saved r8 of caller function
 54136        72            saved r9 of caller function
 54240        80            saved r10 of caller function
 54344        88            saved r11 of caller function
 54448        96            saved r12 of caller function
 54552        104           saved r13 of caller function
 54656        112           saved r14 of caller function
 54760        120           saved r15 of caller function
 54864        128           saved f4 of caller function
 54972        132           saved f6 of caller function
 55080                      undefined
 55196        160           outgoing args passed from caller to callee
 55296+x      160+x         possible stack alignment ( 8 bytes desirable )
 55396+x+y    160+x+y       alloca space of caller ( if used )
 55496+x+y+z  160+x+y+z     automatics of caller ( if used )
 5550                       back-chain
 557A sample program with comments.
 560Comments on the function test
 5621) It didn't need to set up a pointer to the constant pool gpr13 as it isn't used
 563( :-( ).
 5642) This is a frameless function & no stack is bought.
 5653) The compiler was clever enough to recognise that it could return the
 566value in r2 as well as use it for the passed in parameter ( :-) ).
 5674) The basr ( branch relative & save ) trick works as follows the instruction 
 568has a special case with r0,r0 with some instruction operands is understood as 
 569the literal value 0, some risc architectures also do this ). So now
 570we are branching to the next address & the address new program counter is
 571in r13,so now we subtract the size of the function prologue we have executed
 572+ the size of the literal pool to get to the top of the literal pool
 5730040037c int test(int b)
 574{                                                          # Function prologue below
 575  40037c:       90 de f0 34     stm     %r13,%r14,52(%r15) # Save registers r13 & r14
 576  400380:       0d d0           basr    %r13,%r0           # Set up pointer to constant pool using
 577  400382:       a7 da ff fa     ahi     %r13,-6            # basr trick
 578        return(5+b);
 579                                                           # Huge main program
 580  400386:       a7 2a 00 05     ahi     %r2,5              # add 5 to r2
 582                                                           # Function epilogue below 
 583  40038a:       98 de f0 34     lm      %r13,%r14,52(%r15) # restore registers r13 & 14
 584  40038e:       07 fe           br      %r14               # return
 587Comments on the function main
 5891) The compiler did this function optimally ( 8-) )
 591Literal pool for main.
 592400390: ff ff ff ec     .long 0xffffffec
 593main(int argc,char *argv[])
 594{                                                          # Function prologue below
 595  400394:       90 bf f0 2c     stm     %r11,%r15,44(%r15) # Save necessary registers
 596  400398:       18 0f           lr      %r0,%r15           # copy stack pointer to r0
 597  40039a:       a7 fa ff a0     ahi     %r15,-96           # Make area for callee saving 
 598  40039e:       0d d0           basr    %r13,%r0           # Set up r13 to point to
 599  4003a0:       a7 da ff f0     ahi     %r13,-16           # literal pool
 600  4003a4:       50 00 f0 00     st      %r0,0(%r15)        # Save backchain
 602        return(test(5));                                   # Main Program Below
 603  4003a8:       58 e0 d0 00     l       %r14,0(%r13)       # load relative address of test from
 604                                                           # literal pool
 605  4003ac:       a7 28 00 05     lhi     %r2,5              # Set first parameter to 5
 606  4003b0:       4d ee d0 00     bas     %r14,0(%r14,%r13)  # jump to test setting r14 as return
 607                                                           # address using branch & save instruction.
 609                                                           # Function Epilogue below
 610  4003b4:       98 bf f0 8c     lm      %r11,%r15,140(%r15)# Restore necessary registers.
 611  4003b8:       07 fe           br      %r14               # return to do program exit 
 615Compiler updates
 618main(int argc,char *argv[])
 620  4004fc:       90 7f f0 1c             stm     %r7,%r15,28(%r15)
 621  400500:       a7 d5 00 04             bras    %r13,400508 <main+0xc>
 622  400504:       00 40 04 f4             .long   0x004004f4 
 623  # compiler now puts constant pool in code to so it saves an instruction 
 624  400508:       18 0f                   lr      %r0,%r15
 625  40050a:       a7 fa ff a0             ahi     %r15,-96
 626  40050e:       50 00 f0 00             st      %r0,0(%r15)
 627        return(test(5));
 628  400512:       58 10 d0 00             l       %r1,0(%r13)
 629  400516:       a7 28 00 05             lhi     %r2,5
 630  40051a:       0d e1                   basr    %r14,%r1
 631  # compiler adds 1 extra instruction to epilogue this is done to
 632  # avoid processor pipeline stalls owing to data dependencies on g5 &
 633  # above as register 14 in the old code was needed directly after being loaded 
 634  # by the lm   %r11,%r15,140(%r15) for the br %14.
 635  40051c:       58 40 f0 98             l       %r4,152(%r15)
 636  400520:       98 7f f0 7c             lm      %r7,%r15,124(%r15)
 637  400524:       07 f4                   br      %r4
 641Hartmut ( our compiler developer ) also has been threatening to take out the
 642stack backchain in optimised code as this also causes pipeline stalls, you
 643have been warned.
 64564 bit z/Architecture code disassembly
 648If you understand the stuff above you'll understand the stuff
 649below too so I'll avoid repeating myself & just say that 
 650some of the instructions have g's on the end of them to indicate
 651they are 64 bit & the stack offsets are a bigger, 
 652the only other difference you'll find between 32 & 64 bit is that
 653we now use f4 & f6 for floating point arguments on 64 bit.
 65400000000800005b0 <test>:
 655int test(int b)
 657        return(5+b);
 658    800005b0:   a7 2a 00 05             ahi     %r2,5
 659    800005b4:   b9 14 00 22             lgfr    %r2,%r2 # downcast to integer
 660    800005b8:   07 fe                   br      %r14
 661    800005ba:   07 07                   bcr     0,%r7
 66600000000800005bc <main>:
 667main(int argc,char *argv[])
 669    800005bc:   eb bf f0 58 00 24       stmg    %r11,%r15,88(%r15)
 670    800005c2:   b9 04 00 1f             lgr     %r1,%r15
 671    800005c6:   a7 fb ff 60             aghi    %r15,-160
 672    800005ca:   e3 10 f0 00 00 24       stg     %r1,0(%r15)
 673        return(test(5));
 674    800005d0:   a7 29 00 05             lghi    %r2,5
 675    # brasl allows jumps > 64k & is overkill here bras would do fune
 676    800005d4:   c0 e5 ff ff ff ee       brasl   %r14,800005b0 <test> 
 677    800005da:   e3 40 f1 10 00 04       lg      %r4,272(%r15)
 678    800005e0:   eb bf f0 f8 00 04       lmg     %r11,%r15,248(%r15)
 679    800005e6:   07 f4                   br      %r4
 684Compiling programs for debugging on Linux for s/390 & z/Architecture
 686-gdwarf-2 now works it should be considered the default debugging
 687format for s/390 & z/Architecture as it is more reliable for debugging
 688shared libraries,  normal -g debugging works much better now
 689Thanks to the IBM java compiler developers bug reports. 
 691This is typically done adding/appending the flags -g or -gdwarf-2 to the 
 692CFLAGS & LDFLAGS variables Makefile of the program concerned.
 694If using gdb & you would like accurate displays of registers &
 695 stack traces compile without optimisation i.e make sure
 696that there is no -O2 or similar on the CFLAGS line of the Makefile &
 697the emitted gcc commands, obviously this will produce worse code 
 698( not advisable for shipment ) but it is an  aid to the debugging process.
 700This aids debugging because the compiler will copy parameters passed in
 701in registers onto the stack so backtracing & looking at passed in
 702parameters will work, however some larger programs which use inline functions
 703will not compile without optimisation.
 705Debugging with optimisation has since much improved after fixing
 706some bugs, please make sure you are using gdb-5.0 or later developed 
 707after Nov'2000.
 709Figuring out gcc compile errors
 711If you are getting a lot of syntax errors compiling a program & the problem
 712isn't blatantly obvious from the source.
 713It often helps to just preprocess the file, this is done with the -E
 714option in gcc.
 715What this does is that it runs through the very first phase of compilation
 716( compilation in gcc is done in several stages & gcc calls many programs to
 717achieve its end result ) with the -E option gcc just calls the gcc preprocessor (cpp).
 718The c preprocessor does the following, it joins all the files #included together
 719recursively ( #include files can #include other files ) & also the c file you wish to compile.
 720It puts a fully qualified path of the #included files in a comment & it
 721does macro expansion.
 722This is useful for debugging because
 7231) You can double check whether the files you expect to be included are the ones
 724that are being included ( e.g. double check that you aren't going to the i386 asm directory ).
 7252) Check that macro definitions aren't clashing with typedefs,
 7263) Check that definitions aren't being used before they are being included.
 7274) Helps put the line emitting the error under the microscope if it contains macros.
 729For convenience the Linux kernel's makefile will do preprocessing automatically for you
 730by suffixing the file you want built with .i ( instead of .o )
 733from the linux directory type
 734make arch/s390/kernel/signal.i
 735this will build
 737s390-gcc -D__KERNEL__ -I/home1/barrow/linux/include -Wall -Wstrict-prototypes -O2 -fomit-frame-pointer
 738-fno-strict-aliasing -D__SMP__ -pipe -fno-strength-reduce   -E arch/s390/kernel/signal.c
 739> arch/s390/kernel/signal.i  
 741Now look at signal.i you should see something like.
 744# 1 "/home1/barrow/linux/include/asm/types.h" 1
 745typedef unsigned short umode_t;
 746typedef __signed__ char __s8;
 747typedef unsigned char __u8;
 748typedef __signed__ short __s16;
 749typedef unsigned short __u16;
 751If instead you are getting errors further down e.g.
 752unknown instruction:2515 "move.l" or better still unknown instruction:2515 
 753"Fixme not implemented yet, call Martin" you are probably are attempting to compile some code 
 754meant for another architecture or code that is simply not implemented, with a fixme statement
 755stuck into the inline assembly code so that the author of the file now knows he has work to do.
 756To look at the assembly emitted by gcc just before it is about to call gas ( the gnu assembler )
 757use the -S option.
 758Again for your convenience the Linux kernel's Makefile will hold your hand &
 759do all this donkey work for you also by building the file with the .s suffix.
 761from the Linux directory type 
 762make arch/s390/kernel/signal.s 
 764s390-gcc -D__KERNEL__ -I/home1/barrow/linux/include -Wall -Wstrict-prototypes -O2 -fomit-frame-pointer
 765-fno-strict-aliasing -D__SMP__ -pipe -fno-strength-reduce  -S arch/s390/kernel/signal.c 
 766-o arch/s390/kernel/signal.s  
 769This will output something like, ( please note the constant pool & the useful comments
 770in the prologue to give you a hand at interpreting it ).
 773        .string "misaligned (__u16 *) in __xchg\n"
 775        .string "misaligned (__u32 *) in __xchg\n"
 776.L$PG1: # Pool sys_sigsuspend
 778        .long   -262401
 780        .long   -1
 782        .long   schedule-.L$PG1
 784        .long   do_signal-.L$PG1
 785        .align 4
 786.globl sys_sigsuspend
 787        .type    sys_sigsuspend,@function
 789#       leaf function           0
 790#       automatics              16
 791#       outgoing args           0
 792#       need frame pointer      0
 793#       call alloca             0
 794#       has varargs             0
 795#       incoming args (stack)   0
 796#       function length         168
 797        STM     8,15,32(15)
 798        LR      0,15
 799        AHI     15,-112
 800        BASR    13,0
 801.L$CO1: AHI     13,.L$PG1-.L$CO1
 802        ST      0,0(15)
 803        LR    8,2
 804        N     5,.LC192-.L$PG1(13) 
 806Adding -g to the above output makes the output even more useful
 807e.g. typing
 808make CC:="s390-gcc -g" kernel/sched.s
 810which compiles.
 811s390-gcc -g -D__KERNEL__ -I/home/barrow/linux-2.3/include -Wall -Wstrict-prototypes -O2 -fomit-frame-pointer -fno-strict-aliasing -pipe -fno-strength-reduce   -S kernel/sched.c -o kernel/sched.s 
 813also outputs stabs ( debugger ) info, from this info you can find out the
 814offsets & sizes of various elements in structures.
 815e.g. the stab for the structure
 816struct rlimit {
 817        unsigned long   rlim_cur;
 818        unsigned long   rlim_max;
 821.stabs "rlimit:T(151,2)=s8rlim_cur:(0,5),0,32;rlim_max:(0,5),32,32;;",128,0,0,0
 822from this stab you can see that 
 823rlimit_cur starts at bit offset 0 & is 32 bits in size
 824rlimit_max starts at bit offset 32 & is 32 bits in size.
 827Debugging Tools:
 832This is a tool with many options the most useful being ( if compiled with -g).
 833objdump --source <victim program or object file> > <victims debug listing >
 836The whole kernel can be compiled like this ( Doing this will make a 17MB kernel
 837& a 200 MB listing ) however you have to strip it before building the image
 838using the strip command to make it a more reasonable size to boot it.
 840A source/assembly mixed dump of the kernel can be done with the line
 841objdump --source vmlinux > vmlinux.lst
 842Also, if the file isn't compiled -g, this will output as much debugging information
 843as it can (e.g. function names). This is very slow as it spends lots
 844of time searching for debugging info. The following self explanatory line should be used 
 845instead if the code isn't compiled -g, as it is much faster:
 846objdump --disassemble-all --syms vmlinux > vmlinux.lst  
 848As hard drive space is valuable most of us use the following approach.
 8491) Look at the emitted psw on the console to find the crash address in the kernel.
 8502) Look at the file ( in the linux directory ) produced when building 
 851the kernel to find the closest address less than the current PSW to find the
 852offending function.
 8533) use grep or similar to search the source tree looking for the source file
 854 with this function if you don't know where it is.
 8554) rebuild this object file with -g on, as an example suppose the file was
 856( /arch/s390/kernel/signal.o ) 
 8575) Assuming the file with the erroneous function is signal.c Move to the base of the 
 858Linux source tree.
 8596) rm /arch/s390/kernel/signal.o
 8607) make /arch/s390/kernel/signal.o
 8618) watch the gcc command line emitted
 8629) type it in again or alternatively cut & paste it on the console adding the -g option.
 86310) objdump --source arch/s390/kernel/signal.o > signal.lst
 864This will output the source & the assembly intermixed, as the snippet below shows
 865This will unfortunately output addresses which aren't the same
 866as the kernel ones you should be able to get around the mental arithmetic
 867by playing with the --adjust-vma parameter to objdump.
 872static inline void spin_lock(spinlock_t *lp)
 874      a0:       18 34           lr      %r3,%r4
 875      a2:       a7 3a 03 bc     ahi     %r3,956
 876        __asm__ __volatile("    lhi   1,-1\n"
 877      a6:       a7 18 ff ff     lhi     %r1,-1
 878      aa:       1f 00           slr     %r0,%r0
 879      ac:       ba 01 30 00     cs      %r0,%r1,0(%r3)
 880      b0:       a7 44 ff fd     jm      aa <sys_sigsuspend+0x2e>
 881        saveset = current->blocked;
 882      b4:       d2 07 f0 68     mvc     104(8,%r15),972(%r4)
 883      b8:       43 cc
 884        return (set->sig[0] & mask) != 0;
 8876) If debugging under VM go down to that section in the document for more info.
 890I now have a tool which takes the pain out of --adjust-vma
 891& you are able to do something like
 892make /arch/s390/kernel/traps.lst
 893& it automatically generates the correctly relocated entries for
 894the text segment in traps.lst.
 895This tool is now standard in linux distro's in scripts/makelst
 899Q. What is it ?
 900A. It is a tool for intercepting calls to the kernel & logging them
 901to a file & on the screen.
 903Q. What use is it ?
 904A. You can use it to find out what files a particular program opens.
 908Example 1
 910If you wanted to know does ping work but didn't have the source 
 911strace ping -c 1  
 912& then look at the man pages for each of the syscalls below,
 913( In fact this is sometimes easier than looking at some spaghetti
 914source which conditionally compiles for several architectures ).
 915Not everything that it throws out needs to make sense immediately.
 917Just looking quickly you can see that it is making up a RAW socket
 918for the ICMP protocol.
 919Doing an alarm(10) for a 10 second timeout
 920& doing a gettimeofday call before & after each read to see 
 921how long the replies took, & writing some text to stdout so the user
 922has an idea what is going on.
 925getuid()                                = 0
 926setuid(0)                               = 0
 927stat("/usr/share/locale/C/", 0xbffff134) = -1 ENOENT (No such file or directory)
 928stat("/usr/share/locale/libc/C", 0xbffff134) = -1 ENOENT (No such file or directory)
 929stat("/usr/local/share/locale/C/", 0xbffff134) = -1 ENOENT (No such file or directory)
 930getpid()                                = 353
 931setsockopt(3, SOL_SOCKET, SO_BROADCAST, [1], 4) = 0
 932setsockopt(3, SOL_SOCKET, SO_RCVBUF, [49152], 4) = 0
 933fstat(1, {st_mode=S_IFCHR|0620, st_rdev=makedev(3, 1), ...}) = 0
 934mmap(0, 4096, PROT_READ|PROT_WRITE, MAP_PRIVATE|MAP_ANONYMOUS, -1, 0) = 0x40008000
 935ioctl(1, TCGETS, {B9600 opost isig icanon echo ...}) = 0
 936write(1, "PING ( 56 d"..., 42PING ( 56 data bytes
 937) = 42
 938sigaction(SIGINT, {0x8049ba0, [], SA_RESTART}, {SIG_DFL}) = 0 
 939sigaction(SIGALRM, {0x8049600, [], SA_RESTART}, {SIG_DFL}) = 0
 940gettimeofday({948904719, 138951}, NULL) = 0
 941sendto(3, "\10\0D\201a\1\0\0\17#\2178\307\36"..., 64, 0, {sin_family=AF_INET,
 942sin_port=htons(0), sin_addr=inet_addr("")}, 16) = 64
 943sigaction(SIGALRM, {0x8049600, [], SA_RESTART}, {0x8049600, [], SA_RESTART}) = 0
 944sigaction(SIGALRM, {0x8049ba0, [], SA_RESTART}, {0x8049600, [], SA_RESTART}) = 0
 945alarm(10)                               = 0
 946recvfrom(3, "E\0\0T\0005\0\0@\1|r\177\0\0\1\177"..., 192, 0, 
 947{sin_family=AF_INET, sin_port=htons(50882), sin_addr=inet_addr("")}, [16]) = 84
 948gettimeofday({948904719, 160224}, NULL) = 0
 949recvfrom(3, "E\0\0T\0006\0\0\377\1\275p\177\0"..., 192, 0, 
 950{sin_family=AF_INET, sin_port=htons(50882), sin_addr=inet_addr("")}, [16]) = 84
 951gettimeofday({948904719, 166952}, NULL) = 0
 952write(1, "64 bytes from icmp_se"..., 
 9535764 bytes from icmp_seq=0 ttl=255 time=28.0 ms
 955Example 2
 957strace passwd 2>&1 | grep open
 958produces the following output
 959open("/etc/", O_RDONLY)      = 3
 960open("/opt/kde/lib/", O_RDONLY) = -1 ENOENT (No such file or directory)
 961open("/lib/", O_RDONLY)        = 3
 962open("/dev", O_RDONLY)                  = 3
 963open("/var/run/utmp", O_RDONLY)         = 3
 964open("/etc/passwd", O_RDONLY)           = 3
 965open("/etc/shadow", O_RDONLY)           = 3
 966open("/etc/login.defs", O_RDONLY)       = 4
 967open("/dev/tty", O_RDONLY)              = 4 
 969The 2>&1 is done to redirect stderr to stdout & grep is then filtering this input 
 970through the pipe for each line containing the string open.
 973Example 3
 975Getting sophisticated
 976telnetd crashes & I don't know why
 9801) Replace the following line in /etc/inetd.conf
 981telnet  stream  tcp     nowait  root    /usr/sbin/in.telnetd -h 
 983telnet  stream  tcp     nowait  root    /blah
 9852) Create the file /blah with the following contents to start tracing telnetd 
 987/usr/bin/strace -o/t1 -f /usr/sbin/in.telnetd -h 
 9883) chmod 700 /blah to make it executable only to root
 990killall -HUP inetd
 991or ps aux | grep inetd
 992get inetd's process id
 993& kill -HUP inetd to restart it.
 995Important options
 997-o is used to tell strace to output to a file in our case t1 in the root directory
 998-f is to follow children i.e.
 999e.g in our case above telnetd will start the login process & subsequently a shell like bash.
1000You will be able to tell which is which from the process ID's listed on the left hand side
1001of the strace output.
1002-p<pid> will tell strace to attach to a running process, yup this can be done provided
1003 it isn't being traced or debugged already & you have enough privileges,
1004the reason 2 processes cannot trace or debug the same program is that strace
1005becomes the parent process of the one being debugged & processes ( unlike people )
1006can have only one parent.
1009However the file /t1 will get big quite quickly
1010to test it telnet
1012now look at what files in.telnetd execve'd
1013413   execve("/usr/sbin/in.telnetd", ["/usr/sbin/in.telnetd", "-h"], [/* 17 vars */]) = 0
1014414   execve("/bin/login", ["/bin/login", "-h", "localhost", "-p"], [/* 2 vars */]) = 0 
1016Whey it worked!.
1019Other hints:
1021If the program is not very interactive ( i.e. not much keyboard input )
1022& is crashing in one architecture but not in another you can do 
1023an strace of both programs under as identical a scenario as you can
1024on both architectures outputting to a file then.
1025do a diff of the two traces using the diff program
1027diff output1 output2
1028& maybe you'll be able to see where the call paths differed, this
1029is possibly near the cause of the crash. 
1031More info
1033Look at man pages for strace & the various syscalls
1034e.g. man strace, man alarm, man socket.
1037Performance Debugging
1039gcc is capable of compiling in profiling code just add the -p option
1040to the CFLAGS, this obviously affects program size & performance.
1041This can be used by the gprof gnu profiling tool or the
1042gcov the gnu code coverage tool ( code coverage is a means of testing
1043code quality by checking if all the code in an executable in exercised by
1044a tester ).
1047Using top to find out where processes are sleeping in the kernel
1049To do this copy the from the root directory where
1050the linux kernel was built to the /boot directory on your 
1051linux machine.
1052Start top
1053Now type fU<return>
1054You should see a new field called WCHAN which
1055tells you where each process is sleeping here is a typical output.
1057 6:59pm  up 41 min,  1 user,  load average: 0.00, 0.00, 0.00
105828 processes: 27 sleeping, 1 running, 0 zombie, 0 stopped
1059CPU states:  0.0% user,  0.1% system,  0.0% nice, 99.8% idle
1060Mem:   254900K av,   45976K used,  208924K free,       0K shrd,   28636K buff
1061Swap:       0K av,       0K used,       0K free                    8620K cached
1064  750 root      12   0   848  848   700 do_select S       0  0.1  0.3   0:00 in.telnetd
1065  767 root      16   0  1140 1140   964           R       0  0.1  0.4   0:00 top
1066    1 root       8   0   212  212   180 do_select S       0  0.0  0.0   0:00 init
1067    2 root       9   0     0    0     0 down_inte SW      0  0.0  0.0   0:00 kmcheck
1069The time command
1071Another related command is the time command which gives you an indication
1072of where a process is spending the majority of its time.
1074time ping -c 5 nc
1076real    0m4.054s
1077user    0m0.010s
1078sys     0m0.010s
1080Debugging under VM
1085Addresses & values in the VM debugger are always hex never decimal
1086Address ranges are of the format <HexValue1>-<HexValue2> or <HexValue1>.<HexValue2> 
1087e.g. The address range  0x2000 to 0x3000 can be described as 2000-3000 or 2000.1000
1089The VM Debugger is case insensitive.
1091VM's strengths are usually other debuggers weaknesses you can get at any resource
1092no matter how sensitive e.g. memory management resources,change address translation
1093in the PSW. For kernel hacking you will reap dividends if you get good at it.
1095The VM Debugger displays operators but not operands, probably because some
1096of it was written when memory was expensive & the programmer was probably proud that
1097it fitted into 2k of memory & the programmers & didn't want to shock hardcore VM'ers by
1098changing the interface :-), also the debugger displays useful information on the same line & 
1099the author of the code probably felt that it was a good idea not to go over 
1100the 80 columns on the screen. 
1102As some of you are probably in a panic now this isn't as unintuitive as it may seem
1103as the 390 instructions are easy to decode mentally & you can make a good guess at a lot 
1104of them as all the operands are nibble ( half byte aligned ) & if you have an objdump listing
1105also it is quite easy to follow, if you don't have an objdump listing keep a copy of
1106the s/390 Reference Summary & look at between pages 2 & 7 or alternatively the
1107s/390 principles of operation.
1108e.g. even I can guess that 
11090001AFF8' LR    180F        CC 0
1110is a ( load register ) lr r0,r15 
1112Also it is very easy to tell the length of a 390 instruction from the 2 most significant
1113bits in the instruction ( not that this info is really useful except if you are trying to
1114make sense of a hexdump of code ).
1115Here is a table
1116Bits                    Instruction Length
111800                          2 Bytes
111901                          4 Bytes
112010                          4 Bytes
112111                          6 Bytes
1126The debugger also displays other useful info on the same line such as the
1127addresses being operated on destination addresses of branches & condition codes.
112900019736' AHI   A7DAFF0E    CC 1
1130000198BA' BRC   A7840004 -> 000198C2'   CC 0
1131000198CE' STM   900EF068 >> 0FA95E78    CC 2
1135Useful VM debugger commands
1138I suppose I'd better mention this before I start
1139to list the current active traces do 
1140Q TR
1141there can be a maximum of 255 of these per set
1142( more about trace sets later ).
1143To stop traces issue a
1144TR END.
1145To delete a particular breakpoint issue
1146TR DEL <breakpoint number>
1148The PA1 key drops to CP mode so you can issue debugger commands,
1149Doing alt c (on my 3270 console at least ) clears the screen. 
1150hitting b <enter> comes back to the running operating system
1151from cp mode ( in our case linux ).
1152It is typically useful to add shortcuts to your profile.exec file
1153if you have one ( this is roughly equivalent to autoexec.bat in DOS ).
1154file here are a few from mine.
1155/* this gives me command history on issuing f12 */
1156set pf12 retrieve 
1157/* this continues */
1158set pf8 imm b
1159/* goes to trace set a */
1160set pf1 imm tr goto a
1161/* goes to trace set b */
1162set pf2 imm tr goto b
1163/* goes to trace set c */
1164set pf3 imm tr goto c
1168Instruction Tracing
1170Setting a simple breakpoint
1171TR I PSWA <address>
1172To debug a particular function try
1173TR I R <function address range>
1174TR I on its own will single step.
1175TR I DATA <MNEMONIC> <OPTIONAL RANGE> will trace for particular mnemonics
1177TR I DATA 4D R 0197BC.4000
1178will trace for BAS'es ( opcode 4D ) in the range 0197BC.4000
1179if you were inclined you could add traces for all branch instructions &
1180suffix them with the run prefix so you would have a backtrace on screen 
1181when a program crashes.
1182TR BR <INTO OR FROM> will trace branches into or out of an address.
1184TR BR INTO 0 is often quite useful if a program is getting awkward & deciding
1185to branch to 0 & crashing as this will stop at the address before in jumps to 0.
1186TR I R <address range> RUN cmd d g
1187single steps a range of addresses but stays running &
1188displays the gprs on each step.
1192Displaying & modifying Registers
1194D G will display all the gprs
1195Adding a extra G to all the commands is necessary to access the full 64 bit 
1196content in VM on z/Architecture obviously this isn't required for access registers
1197as these are still 32 bit.
1198e.g. DGG instead of DG 
1199D X will display all the control registers
1200D AR will display all the access registers
1201D AR4-7 will display access registers 4 to 7
1202CPU ALL D G will display the GRPS of all CPUS in the configuration
1203D PSW will display the current PSW
1204st PSW 2000 will put the value 2000 into the PSW &
1205cause crash your machine.
1206D PREFIX displays the prefix offset
1209Displaying Memory
1211To display memory mapped using the current PSW's mapping try
1212D <range>
1213To make VM display a message each time it hits a particular address & continue try
1214D I<range> will disassemble/display a range of instructions.
1215ST addr 32 bit word will store a 32 bit aligned address
1216D T<range> will display the EBCDIC in an address ( if you are that way inclined )
1217D R<range> will display real addresses ( without DAT ) but with prefixing.
1218There are other complex options to display if you need to get at say home space
1219but are in primary space the easiest thing to do is to temporarily
1220modify the PSW to the other addressing mode, display the stuff & then
1221restore it.
1227If you want to issue a debugger command without halting your virtual machine with the
1228PA1 key try prefixing the command with #CP e.g.
1229#cp tr i pswa 2000
1230also suffixing most debugger commands with RUN will cause them not
1231to stop just display the mnemonic at the current instruction on the console.
1232If you have several breakpoints you want to put into your program &
1233you get fed up of cross referencing with
1234you can do the following trick for several symbols.
1235grep do_signal 
1236which emits the following among other things
12370001f4e0 T do_signal 
1238now you can do
1240TR I PSWA 0001f4e0 cmd msg * do_signal
1241This sends a message to your own console each time do_signal is entered.
1242( As an aside I wrote a perl script once which automatically generated a REXX
1243script with breakpoints on every kernel procedure, this isn't a good idea
1244because there are thousands of these routines & VM can only set 255 breakpoints
1245at a time so you nearly had to spend as long pruning the file down as you would 
1246entering the msg's by hand ),however, the trick might be useful for a single object file.
1247On linux'es 3270 emulator x3270 there is a very useful option under the file ment
1248Save Screens In File this is very good of keeping a copy of traces. 
1250From CMS help <command name> will give you online help on a particular command. 
1254Also CP has a file called profile.exec which automatically gets called
1255on startup of CMS ( like autoexec.bat ), keeping on a DOS analogy session
1256CP has a feature similar to doskey, it may be useful for you to
1257use profile.exec to define some keystrokes. 
1260This does a single step in VM on pressing F8. 
1261SET PF10  ^
1262This sets up the ^ key.
1263which can be used for ^c (ctrl-c),^z (ctrl-z) which can't be typed directly into some 3270 consoles.
1264SET PF11 ^-
1265This types the starting keystrokes for a sysrq see SysRq below.
1267This retrieves command history on pressing F12.
1270Sometimes in VM the display is set up to scroll automatically this
1271can be very annoying if there are messages you wish to look at
1272to stop this do
1273TERM MORE 255 255
1274This will nearly stop automatic screen updates, however it will
1275cause a denial of service if lots of messages go to the 3270 console,
1276so it would be foolish to use this as the default on a production machine.
1279Tracing particular processes
1281The kernel's text segment is intentionally at an address in memory that it will
1282very seldom collide with text segments of user programs ( thanks Martin ),
1283this simplifies debugging the kernel.
1284However it is quite common for user processes to have addresses which collide
1285this can make debugging a particular process under VM painful under normal
1286circumstances as the process may change when doing a 
1287TR I R <address range>.
1288Thankfully after reading VM's online help I figured out how to debug
1289I particular process.
1291Your first problem is to find the STD ( segment table designation )
1292of the program you wish to debug.
1293There are several ways you can do this here are a few
12941) objdump --syms <program to be debugged> | grep main
1295To get the address of main in the program.
1296tr i pswa <address of main>
1297Start the program, if VM drops to CP on what looks like the entry
1298point of the main function this is most likely the process you wish to debug.
1299Now do a D X13 or D XG13 on z/Architecture.
1300On 31 bit the STD is bits 1-19 ( the STO segment table origin ) 
1301& 25-31 ( the STL segment table length ) of CR13.
1302now type
1303TR I R STD <CR13's value> 0.7fffffff
1305TR I R STD 8F32E1FF 0.7fffffff
1306Another very useful variation is
1307TR STORE INTO STD <CR13's value> <address range>
1308for finding out when a particular variable changes.
1310An alternative way of finding the STD of a currently running process 
1311is to do the following, ( this method is more complex but
1312could be quite convenient if you aren't updating the kernel much &
1313so your kernel structures will stay constant for a reasonable period of
1314time ).
1316grep task /proc/<pid>/status
1317from this you should see something like
1318task: 0f160000 ksp: 0f161de8 pt_regs: 0f161f68
1319This now gives you a pointer to the task structure.
1320Now make CC:="s390-gcc -g" kernel/sched.s
1321To get the task_struct stabinfo.
1322( task_struct is defined in include/linux/sched.h ).
1323Now we want to look at
1325on my machine the active_mm in the task structure stab is
1327its offset is 672/8=84=0x54
1328the pgd member in the mm_struct stab is
1330so its offset is 96/8=12=0xc
1332so we'll
1333hexdump -s 0xf160054 /dev/mem | more
1334i.e. task_struct+active_mm offset
1335to look at the active_mm member
1336f160054 0fee cc60 0019 e334 0000 0000 0000 0011
1337hexdump -s 0x0feecc6c /dev/mem | more
1338i.e. active_mm+pgd offset
1339feecc6c 0f2c 0000 0000 0001 0000 0001 0000 0010
1340we get something like
1341now do 
1342TR I R STD <pgd|0x7f> 0.7fffffff
1343i.e. the 0x7f is added because the pgd only
1344gives the page table origin & we need to set the low bits
1345to the maximum possible segment table length.
1346TR I R STD 0f2c007f 0.7fffffff
1347on z/Architecture you'll probably need to do
1348TR I R STD <pgd|0x7> 0.ffffffffffffffff
1349to set the TableType to 0x1 & the Table length to 3.
1353Tracing Program Exceptions
1355If you get a crash which says something like
1356illegal operation or specification exception followed by a register dump
1357You can restart linux & trace these using the tr prog <range or value> trace option.
1361The most common ones you will normally be tracing for is
13621=operation exception
13632=privileged operation exception
13644=protection exception
13655=addressing exception
13666=specification exception
136710=segment translation exception
136811=page translation exception
1370The full list of these is on page 22 of the current s/390 Reference Summary.
1372tr prog 10 will trace segment translation exceptions.
1373tr prog on its own will trace all program interruption codes.
1375Trace Sets
1377On starting VM you are initially in the INITIAL trace set.
1378You can do a Q TR to verify this.
1379If you have a complex tracing situation where you wish to wait for instance 
1380till a driver is open before you start tracing IO, but know in your
1381heart that you are going to have to make several runs through the code till you
1382have a clue whats going on. 
1384What you can do is
1385TR I PSWA <Driver open address>
1386hit b to continue till breakpoint
1387reach the breakpoint
1388now do your
1389TR GOTO B 
1390TR IO 7c08-7c09 inst int run 
1391or whatever the IO channels you wish to trace are & hit b
1393To got back to the initial trace set do
1395& the TR I PSWA <Driver open address> will be the only active breakpoint again.
1398Tracing linux syscalls under VM
1400Syscalls are implemented on Linux for S390 by the Supervisor call instruction (SVC) there 256 
1401possibilities of these as the instruction is made up of a  0xA opcode & the second byte being
1402the syscall number. They are traced using the simple command.
1403TR SVC  <Optional value or range>
1404the syscalls are defined in linux/arch/s390/include/asm/unistd.h
1405e.g. to trace all file opens just do
1406TR SVC 5 ( as this is the syscall number of open )
1409SMP Specific commands
1411To find out how many cpus you have
1412Q CPUS displays all the CPU's available to your virtual machine
1413To find the cpu that the current cpu VM debugger commands are being directed at do
1414Q CPU to change the current cpu VM debugger commands are being directed at do
1415CPU <desired cpu no>
1417On a SMP guest issue a command to all CPUs try prefixing the command with cpu all.
1418To issue a command to a particular cpu try cpu <cpu number> e.g.
1419CPU 01 TR I R 2000.3000
1420If you are running on a guest with several cpus & you have a IO related problem
1421& cannot follow the flow of code but you know it isn't smp related.
1422from the bash prompt issue
1423shutdown -h now or halt.
1424do a Q CPUS to find out how many cpus you have
1425detach each one of them from cp except cpu 0 
1426by issuing a 
1427DETACH CPU 01-(number of cpus in configuration)
1428& boot linux again.
1429TR SIGP will trace inter processor signal processor instructions.
1430DEFINE CPU 01-(number in configuration) 
1431will get your guests cpus back.
1434Help for displaying ascii textstrings
1436On the very latest VM Nucleus'es VM can now display ascii
1437( thanks Neale for the hint ) by doing
1438D TX<lowaddr>.<len>
1440D TX0.100
1444Under older VM debuggers ( I love EBDIC too ) you can use this little program I wrote which
1445will convert a command line of hex digits to ascii text which can be compiled under linux & 
1446you can copy the hex digits from your x3270 terminal to your xterm if you are debugging
1447from a linuxbox.
1449This is quite useful when looking at a parameter passed in as a text string
1450under VM ( unless you are good at decoding ASCII in your head ).
1452e.g. consider tracing an open syscall
1453TR SVC 5
1454We have stopped at a breakpoint
1455000151B0' SVC   0A05     -> 0001909A'   CC 0
1457D 20.8 to check the SVC old psw in the prefix area & see was it from userspace
1458( for the layout of the prefix area consult P18 of the s/390 390 Reference Summary 
1459if you have it available ).
1460V00000020  070C2000 800151B2
1461The problem state bit wasn't set &  it's also too early in the boot sequence
1462for it to be a userspace SVC if it was we would have to temporarily switch the 
1463psw to user space addressing so we could get at the first parameter of the open in
1465Next do a 
1466D G2
1467GPR  2 =  00014CB4
1468Now display what gpr2 is pointing to
1469D 00014CB4.20
1470V00014CB4  2F646576 2F636F6E 736F6C65 00001BF5
1471V00014CC4  FC00014C B4001001 E0001000 B8070707
1472Now copy the text till the first 00 hex ( which is the end of the string
1473to an xterm & do hex2ascii on it.
1474hex2ascii 2F646576 2F636F6E 736F6C65 00 
1476Decoded Hex:=/ d e v / c o n s o l e 0x00 
1477We were opening the console device,
1479You can compile the code below yourself for practice :-),
1481 *    hex2ascii.c
1482 *    a useful little tool for converting a hexadecimal command line to ascii
1483 *
1484 *    Author(s): Denis Joseph Barrow (,
1485 *    (C) 2000 IBM Deutschland Entwicklung GmbH, IBM Corporation.
1486 */   
1487#include <stdio.h>
1489int main(int argc,char *argv[])
1491  int cnt1,cnt2,len,toggle=0;
1492  int startcnt=1;
1493  unsigned char c,hex;
1495  if(argc>1&&(strcmp(argv[1],"-a")==0))
1496     startcnt=2;
1497  printf("Decoded Hex:=");
1498  for(cnt1=startcnt;cnt1<argc;cnt1++)
1499  {
1500    len=strlen(argv[cnt1]);
1501    for(cnt2=0;cnt2<len;cnt2++)
1502    {
1503       c=argv[cnt1][cnt2];
1504       if(c>='0'&&c<='9')
1505          c=c-'0';
1506       if(c>='A'&&c<='F')
1507          c=c-'A'+10;
1508       if(c>='a'&&c<='f')
1509          c=c-'a'+10;
1510       switch(toggle)
1511       {
1512          case 0:
1513             hex=c<<4;
1514             toggle=1;
1515          break;
1516          case 1:
1517             hex+=c;
1518             if(hex<32||hex>127)
1519             {
1520                if(startcnt==1)
1521                   printf("0x%02X ",(int)hex);
1522                else
1523                   printf(".");
1524             }
1525             else
1526             {
1527               printf("%c",hex);
1528               if(startcnt==1)
1529                  printf(" ");
1530             }
1531             toggle=0;
1532          break;
1533       }
1534    }
1535  }
1536  printf("\n");
1542Stack tracing under VM
1544A basic backtrace
1547Here are the tricks I use 9 out of 10 times it works pretty well,
1549When your backchain reaches a dead end
1551This can happen when an exception happens in the kernel & the kernel is entered twice
1552if you reach the NULL pointer at the end of the back chain you should be
1553able to sniff further back if you follow the following tricks.
15541) A kernel address should be easy to recognise since it is in
1555primary space & the problem state bit isn't set & also
1556The Hi bit of the address is set.
15572) Another backchain should also be easy to recognise since it is an 
1558address pointing to another address approximately 100 bytes or 0x70 hex
1559behind the current stackpointer.
1562Here is some practice.
1563boot the kernel & hit PA1 at some random time
1564d g to display the gprs, this should display something like
1565GPR  0 =  00000001  00156018  0014359C  00000000
1566GPR  4 =  00000001  001B8888  000003E0  00000000
1567GPR  8 =  00100080  00100084  00000000  000FE000
1568GPR 12 =  00010400  8001B2DC  8001B36A  000FFED8
1569Note that GPR14 is a return address but as we are real men we are going to
1570trace the stack.
1571display 0x40 bytes after the stack pointer.
1573V000FFED8  000FFF38 8001B838 80014C8E 000FFF38
1574V000FFEE8  00000000 00000000 000003E0 00000000
1575V000FFEF8  00100080 00100084 00000000 000FE000
1576V000FFF08  00010400 8001B2DC 8001B36A 000FFED8
1579Ah now look at whats in sp+56 (sp+0x38) this is 8001B36A our saved r14 if
1580you look above at our stackframe & also agrees with GPR14.
1582now backchain 
1583d 000FFF38.40
1584we now are taking the contents of SP to get our first backchain.
1586V000FFF38  000FFFA0 00000000 00014995 00147094
1587V000FFF48  00147090 001470A0 000003E0 00000000
1588V000FFF58  00100080 00100084 00000000 001BF1D0
1589V000FFF68  00010400 800149BA 80014CA6 000FFF38
1591This displays a 2nd return address of 80014CA6
1593now do d 000FFFA0.40 for our 3rd backchain
1595V000FFFA0  04B52002 0001107F 00000000 00000000
1596V000FFFB0  00000000 00000000 FF000000 0001107F
1597V000FFFC0  00000000 00000000 00000000 00000000
1598V000FFFD0  00010400 80010802 8001085A 000FFFA0
1601our 3rd return address is 8001085A
1603as the 04B52002 looks suspiciously like rubbish it is fair to assume that the kernel entry routines
1604for the sake of optimisation don't set up a backchain.
1606now look at to see if the addresses make any sense.
1608grep -i 0001b3
1609outputs among other things
16100001b304 T cpu_idle 
1611so 8001B36A
1612is cpu_idle+0x66 ( quiet the cpu is asleep, don't wake it )
1615grep -i 00014 
1616produces among other things
161700014a78 T start_kernel  
1618so 0014CA6 is start_kernel+some hex number I can't add in my head.
1620grep -i 00108 
1621this produces
162200010800 T _stext
1623so   8001085A is _stext+0x5a
1625Congrats you've done your first backchain.
1629s/390 & z/Architecture IO Overview
1632I am not going to give a course in 390 IO architecture as this would take me quite a
1633while & I'm no expert. Instead I'll give a 390 IO architecture summary for Dummies if you have 
1634the s/390 principles of operation available read this instead. If nothing else you may find a few 
1635useful keywords in here & be able to use them on a web search engine like altavista to find 
1636more useful information.
1638Unlike other bus architectures modern 390 systems do their IO using mostly
1639fibre optics & devices such as tapes & disks can be shared between several mainframes,
1640also S390 can support up to 65536 devices while a high end PC based system might be choking
1641with around 64. Here is some of the common IO terminology
1644This is the logical number most IO commands use to talk to an IO device there can be up to
16450x10000 (65536) of these in a configuration typically there is a few hundred. Under VM
1646for simplicity they are allocated contiguously, however on the native hardware they are not
1647they typically stay consistent between boots provided no new hardware is inserted or removed.
1648Under Linux for 390 we use these as IRQ's & also when issuing an IO command (CLEAR SUBCHANNEL,
1650TEST SUBCHANNEL ) we use this as the ID of the device we wish to talk to, the most
1651important of these instructions are START SUBCHANNEL ( to start IO ), TEST SUBCHANNEL ( to check
1652whether the IO completed successfully ), & HALT SUBCHANNEL ( to kill IO ), a subchannel
1653can have up to 8 channel paths to a device this offers redundancy if one is not available.
1656Device Number:
1657This number remains static & Is closely tied to the hardware, there are 65536 of these
1658also they are made up of a CHPID ( Channel Path ID, the most significant 8 bits ) 
1659& another lsb 8 bits. These remain static even if more devices are inserted or removed
1660from the hardware, there is a 1 to 1 mapping between Subchannels & Device Numbers provided
1661devices aren't inserted or removed.
1663Channel Control Words:
1664CCWS are linked lists of instructions initially pointed to by an operation request block (ORB),
1665which is initially given to Start Subchannel (SSCH) command along with the subchannel number
1666for the IO subsystem to process while the CPU continues executing normal code.
1667These come in two flavours, Format 0 ( 24 bit for backward )
1668compatibility & Format 1 ( 31 bit ). These are typically used to issue read & write 
1669( & many other instructions ) they consist of a length field & an absolute address field.
1670For each IO typically get 1 or 2 interrupts one for channel end ( primary status ) when the
1671channel is idle & the second for device end ( secondary status ) sometimes you get both
1672concurrently, you check how the IO went on by issuing a TEST SUBCHANNEL at each interrupt,
1673from which you receive an Interruption response block (IRB). If you get channel & device end 
1674status in the IRB without channel checks etc. your IO probably went okay. If you didn't you
1675probably need a doctor to examine the IRB & extended status word etc.
1676If an error occurs, more sophisticated control units have a facility known as
1677concurrent sense this means that if an error occurs Extended sense information will
1678be presented in the Extended status word in the IRB if not you have to issue a
1679subsequent SENSE CCW command after the test subchannel. 
1682TPI( Test pending interrupt) can also be used for polled IO but in multitasking multiprocessor
1683systems it isn't recommended except for checking special cases ( i.e. non looping checks for
1684pending IO etc. ).
1686Store Subchannel & Modify Subchannel can be used to examine & modify operating characteristics
1687of a subchannel ( e.g. channel paths ).
1689Other IO related Terms:
1690Sysplex: S390's Clustering Technology
1691QDIO: S390's new high speed IO architecture to support devices such as gigabit ethernet,
1692this architecture is also designed to be forward compatible with up & coming 64 bit machines.
1695General Concepts 
1697Input Output Processors (IOP's) are responsible for communicating between
1698the mainframe CPU's & the channel & relieve the mainframe CPU's from the
1699burden of communicating with IO devices directly, this allows the CPU's to 
1700concentrate on data processing. 
1702IOP's can use one or more links ( known as channel paths ) to talk to each 
1703IO device. It first checks for path availability & chooses an available one,
1704then starts ( & sometimes terminates IO ).
1705There are two types of channel path: ESCON & the Parallel IO interface.
1707IO devices are attached to control units, control units provide the
1708logic to interface the channel paths & channel path IO protocols to 
1709the IO devices, they can be integrated with the devices or housed separately
1710& often talk to several similar devices ( typical examples would be raid 
1711controllers or a control unit which connects to 1000 3270 terminals ).
1714    +---------------------------------------------------------------+
1715    | +-----+ +-----+ +-----+ +-----+  +----------+  +----------+   |
1716    | | CPU | | CPU | | CPU | | CPU |  |  Main    |  | Expanded |   |
1717    | |     | |     | |     | |     |  |  Memory  |  |  Storage |   |
1718    | +-----+ +-----+ +-----+ +-----+  +----------+  +----------+   | 
1719    |---------------------------------------------------------------+
1720    |   IOP        |      IOP      |       IOP                      |
1721    |---------------------------------------------------------------
1722    | C | C | C | C | C | C | C | C | C | C | C | C | C | C | C | C | 
1723    ----------------------------------------------------------------
1724         ||                                              ||
1725         ||  Bus & Tag Channel Path                      || ESCON
1726         ||  ======================                      || Channel
1727         ||  ||                  ||                      || Path
1728    +----------+               +----------+         +----------+
1729    |          |               |          |         |          |
1730    |    CU    |               |    CU    |         |    CU    |
1731    |          |               |          |         |          |
1732    +----------+               +----------+         +----------+
1733       |      |                     |                |       |
1734+----------+ +----------+      +----------+   +----------+ +----------+
1735|I/O Device| |I/O Device|      |I/O Device|   |I/O Device| |I/O Device|
1736+----------+ +----------+      +----------+   +----------+ +----------+
1737  CPU = Central Processing Unit    
1738  C = Channel                      
1739  IOP = IP Processor               
1740  CU = Control Unit
1742The 390 IO systems come in 2 flavours the current 390 machines support both
1744The Older 360 & 370 Interface,sometimes called the Parallel I/O interface,
1745sometimes called Bus-and Tag & sometimes Original Equipment Manufacturers
1746Interface (OEMI).
1748This byte wide Parallel channel path/bus has parity & data on the "Bus" cable 
1749& control lines on the "Tag" cable. These can operate in byte multiplex mode for
1750sharing between several slow devices or burst mode & monopolize the channel for the
1751whole burst. Up to 256 devices can be addressed  on one of these cables. These cables are
1752about one inch in diameter. The maximum unextended length supported by these cables is
1753125 Meters but this can be extended up to 2km with a fibre optic channel extended 
1754such as a 3044. The maximum burst speed supported is 4.5 megabytes per second however
1755some really old processors support only transfer rates of 3.0, 2.0 & 1.0 MB/sec.
1756One of these paths can be daisy chained to up to 8 control units.
1759ESCON if fibre optic it is also called FICON 
1760Was introduced by IBM in 1990. Has 2 fibre optic cables & uses either leds or lasers
1761for communication at a signaling rate of up to 200 megabits/sec. As 10bits are transferred
1762for every 8 bits info this drops to 160 megabits/sec & to 18.6 Megabytes/sec once
1763control info & CRC are added. ESCON only operates in burst mode.
1765ESCONs typical max cable length is 3km for the led version & 20km for the laser version
1766known as XDF ( extended distance facility ). This can be further extended by using an
1767ESCON director which triples the above mentioned ranges. Unlike Bus & Tag as ESCON is
1768serial it uses a packet switching architecture the standard Bus & Tag control protocol
1769is however present within the packets. Up to 256 devices can be attached to each control
1770unit that uses one of these interfaces.
1772Common 390 Devices include:
1773Network adapters typically OSA2,3172's,2116's & OSA-E gigabit ethernet adapters,
1774Consoles 3270 & 3215 ( a teletype emulated under linux for a line mode console ).
1775DASD's direct access storage devices ( otherwise known as hard disks ).
1776Tape Drives.
1777CTC ( Channel to Channel Adapters ),
1778ESCON or Parallel Cables used as a very high speed serial link
1779between 2 machines. We use 2 cables under linux to do a bi-directional serial link.
1782Debugging IO on s/390 & z/Architecture under VM
1785Now we are ready to go on with IO tracing commands under VM
1787A few self explanatory queries:
1788Q OSA
1789Q CTC
1790Q DISK ( This command is CMS specific )
1791Q DASD
1798Q OSA on my machine returns
1799OSA  7C08 ON OSA   7C08 SUBCHANNEL = 0000
1800OSA  7C09 ON OSA   7C09 SUBCHANNEL = 0001
1801OSA  7C14 ON OSA   7C14 SUBCHANNEL = 0002
1802OSA  7C15 ON OSA   7C15 SUBCHANNEL = 0003
1804If you have a guest with certain privileges you may be able to see devices
1805which don't belong to you. To avoid this, add the option V.
1807Q V OSA
1809Now using the device numbers returned by this command we will
1810Trace the io starting up on the first device 7c08 & 7c09
1811In our simplest case we can trace the 
1812start subchannels
1813like TR SSCH 7C08-7C09
1814or the halt subchannels
1815or TR HSCH 7C08-7C09
1816MSCH's ,STSCH's I think you can guess the rest
1818Ingo's favourite trick is tracing all the IO's & CCWS & spooling them into the reader of another
1819VM guest so he can ftp the logfile back to his own machine.I'll do a small bit of this & give you
1820 a look at the output.
18221) Spool stdout to VM reader
1823SP PRT TO (another vm guest ) or * for the local vm guest
18242) Fill the reader with the trace
1825TR IO 7c08-7c09 INST INT CCW PRT RUN
18263) Start up linux 
1827i 00c  
18284) Finish the trace
1829TR END
18305) close the reader
1831C PRT
18326) list reader contents
18347) copy it to linux4's minidisk 
1835RECEIVE / LOG TXT A1 ( replace
1837filel & press F11 to look at it
1838You should see something like:
184000020942' SSCH  B2334000    0048813C    CC 0    SCH 0000    DEV 7C08
1841          CPA 000FFDF0   PARM 00E2C9C4    KEY 0  FPI C0  LPM 80
1842          CCW    000FFDF0  E4200100 00487FE8   0000  E4240100 ........
1843          IDAL                                      43D8AFE8
1844          IDAL                                      0FB76000
184500020B0A'   I/O DEV 7C08 -> 000197BC'   SCH 0000   PARM 00E2C9C4
184600021628' TSCH  B2354000 >> 00488164    CC 0    SCH 0000    DEV 7C08
1847          CCWA 000FFDF8   DEV STS 0C  SCH STS 00  CNT 00EC
1848           KEY 0   FPI C0  CC 0   CTLS 4007
184900022238' STSCH B2344000 >> 00488108    CC 0    SCH 0000    DEV 7C08
1851If you don't like messing up your readed ( because you possibly booted from it )
1852you can alternatively spool it to another readers guest.
1855Other common VM device related commands
1857These commands are listed only because they have
1858been of use to me in the past & may be of use to
1859you too. For more complete info on each of the commands
1860use type HELP <command> from CMS.
1861detaching devices
1862DET <devno range>
1863ATT <devno range> <guest> 
1864attach a device to guest * for your own guest
1865READY <devno> cause VM to issue a fake interrupt.
1867The VARY command is normally only available to VM administrators.
1868VARY ON PATH <path> TO <devno range>
1869VARY OFF PATH <PATH> FROM <devno range>
1870This is used to switch on or off channel paths to devices.
1872Q CHPID <channel path ID>
1873This displays state of devices using this channel path
1874D SCHIB <subchannel>
1875This displays the subchannel information SCHIB block for the device.
1876this I believe is also only available to administrators.
1877DEFINE CTC <devno>
1878defines a virtual CTC channel to channel connection
18792 need to be defined on each guest for the CTC driver to use.
1880COUPLE  devno userid remote devno
1881Joins a local virtual device to a remote virtual device
1882( commonly used for the CTC driver ).
1884Building a VM ramdisk under CMS which linux can use
1885def vfb-<blocksize> <subchannel> <number blocks>
1886blocksize is commonly 4096 for linux.
1887Formatting it
1888format <subchannel> <driver letter e.g. x> (blksize <blocksize>
1890Sharing a disk between multiple guests
1891LINK userid devno1 devno2 mode password
1895GDB on S390
1897N.B. if compiling for debugging gdb works better without optimisation 
1898( see Compiling programs for debugging )
1902gdb <victim program> <optional corefile>
1904Online help
1906help: gives help on commands
1909help display
1910Note gdb's online help is very good use it.
1915info registers: displays registers other than floating point.
1916info all-registers: displays floating points as well.
1917disassemble: disassembles
1919disassemble without parameters will disassemble the current function
1920disassemble $pc $pc+10 
1922Viewing & modifying variables
1924print or p: displays variable or register
1925e.g. p/x $sp will display the stack pointer
1927display: prints variable or register each time program stops
1929display/x $pc will display the program counter
1930display argc
1932undisplay : undo's display's
1934info breakpoints: shows all current breakpoints
1936info stack: shows stack back trace ( if this doesn't work too well, I'll show you the
1937stacktrace by hand below ).
1939info locals: displays local variables.
1941info args: display current procedure arguments.
1943set args: will set argc & argv each time the victim program is invoked.
1945set <variable>=value
1946set argc=100
1947set $pc=0
1951Modifying execution
1953step: steps n lines of sourcecode
1954step steps 1 line.
1955step 100 steps 100 lines of code.
1957next: like step except this will not step into subroutines
1959stepi: steps a single machine code instruction.
1960e.g. stepi 100
1962nexti: steps a single machine code instruction but will not step into subroutines.
1964finish: will run until exit of the current routine
1966run: (re)starts a program
1968cont: continues a program
1970quit: exits gdb.
1977sets a breakpoint
1980break main
1982break *$pc
1984break *0x400618
1986Here's a really useful one for large programs
1988Set a breakpoint for all functions matching REGEXP
1990rbr 390
1991will set a breakpoint with all functions with 390 in their name.
1993info breakpoints
1994lists all breakpoints
1996delete: delete breakpoint by number or delete them all
1998delete 1 will delete the first breakpoint
1999delete will delete them all
2001watch: This will set a watchpoint ( usually hardware assisted ),
2002This will watch a variable till it changes
2004watch cnt, will watch the variable cnt till it changes.
2005As an aside unfortunately gdb's, architecture independent watchpoint code
2006is inconsistent & not very good, watchpoints usually work but not always.
2008info watchpoints: Display currently active watchpoints
2010condition: ( another useful one )
2011Specify breakpoint number N to break only if COND is true.
2012Usage is `condition N COND', where N is an integer and COND is an
2013expression to be evaluated whenever breakpoint N is reached.
2017User defined functions/macros
2019define: ( Note this is very very useful,simple & powerful )
2020usage define <name> <list of commands> end
2022examples which you should consider putting into .gdbinit in your home directory
2023define d
2025disassemble $pc $pc+10
2028define e
2030disassemble $pc $pc+10
2034Other hard to classify stuff
2036signal n:
2037sends the victim program a signal.
2038e.g. signal 3 will send a SIGQUIT.
2040info signals:
2041what gdb does when the victim receives certain signals.
2045list lists current function source
2046list 1,10 list first 10 lines of current file.
2047list test.c:1,10
2051Adds directories to be searched for source if gdb cannot find the source.
2052(note it is a bit sensitive about slashes)
2053e.g. To add the root of the filesystem to the searchpath do
2054directory //
2057call <function>
2058This calls a function in the victim program, this is pretty powerful
2060(gdb) call printf("hello world")
2062$1 = 11 
2064You might now be thinking that the line above didn't work, something extra had to be done.
2065(gdb) call fflush(stdout)
2066hello world$2 = 0
2067As an aside the debugger also calls malloc & free under the hood 
2068to make space for the "hello world" string.
20741) command completion works just like bash 
2075( if you are a bad typist like me this really helps )
2076e.g. hit br <TAB> & cursor up & down :-).
20782) if you have a debugging problem that takes a few steps to recreate
2079put the steps into a file called .gdbinit in your current working directory
2080if you have defined a few extra useful user defined commands put these in 
2081your home directory & they will be read each time gdb is launched.
2083A typical .gdbinit file might be.
2084break main
2086break runtime_exception
2090stack chaining in gdb by hand
2092This is done using a the same trick described for VM 
2093p/x (*($sp+56))&0x7fffffff get the first backchain.
2095For z/Architecture
2096Replace 56 with 112 & ignore the &0x7fffffff
2097in the macros below & do nasty casts to longs like the following
2098as gdb unfortunately deals with printed arguments as ints which
2099messes up everything.
2100i.e. here is a 3rd backchain dereference
2101p/x *(long *)(***(long ***)$sp+112)
2104this outputs 
2105$5 = 0x528f18 
2106on my machine.
2107Now you can use 
2108info symbol (*($sp+56))&0x7fffffff 
2109you might see something like.
2110rl_getc + 36 in section .text  telling you what is located at address 0x528f18
2111Now do.
2112p/x (*(*$sp+56))&0x7fffffff 
2113This outputs
2114$6 = 0x528ed0
2115Now do.
2116info symbol (*(*$sp+56))&0x7fffffff
2117rl_read_key + 180 in section .text
2118now do
2119p/x (*(**$sp+56))&0x7fffffff
2120& so on.
2122Disassembling instructions without debug info
2124gdb typically complains if there is a lack of debugging
2125symbols in the disassemble command with 
2126"No function contains specified address." To get around
2127this do 
2128x/<number lines to disassemble>xi <address>
2130x/20xi 0x400730
2134Note: Remember gdb has history just like bash you don't need to retype the
2135whole line just use the up & down arrows.
2139For more info
2141From your linuxbox do 
2142man gdb or info gdb.
2144core dumps
2146What a core dump ?,
2147A core dump is a file generated by the kernel ( if allowed ) which contains the registers,
2148& all active pages of the program which has crashed.
2149From this file gdb will allow you to look at the registers & stack trace & memory of the
2150program as if it just crashed on your system, it is usually called core & created in the
2151current working directory.
2152This is very useful in that a customer can mail a core dump to a technical support department
2153& the technical support department can reconstruct what happened.
2154Provided they have an identical copy of this program with debugging symbols compiled in &
2155the source base of this build is available.
2156In short it is far more useful than something like a crash log could ever hope to be.
2158In theory all that is missing to restart a core dumped program is a kernel patch which
2159will do the following.
21601) Make a new kernel task structure
21612) Reload all the dumped pages back into the kernel's memory management structures.
21623) Do the required clock fixups
21634) Get all files & network connections for the process back into an identical state ( really difficult ).
21645) A few more difficult things I haven't thought of.
2168Why have I never seen one ?.
2169Probably because you haven't used the command 
2170ulimit -c unlimited in bash
2171to allow core dumps, now do 
2172ulimit -a 
2173to verify that the limit was accepted.
2175A sample core dump
2176To create this I'm going to do
2177ulimit -c unlimited
2179to launch gdb (my victim app. ) now be bad & do the following from another 
2180telnet/xterm session to the same machine
2181ps -aux | grep gdb
2182kill -SIGSEGV <gdb's pid>
2183or alternatively use killall -SIGSEGV gdb if you have the killall command.
2184Now look at the core dump.
2185./gdb core
2186Displays the following
2187GNU gdb 4.18
2188Copyright 1998 Free Software Foundation, Inc.
2189GDB is free software, covered by the GNU General Public License, and you are
2190welcome to change it and/or distribute copies of it under certain conditions.
2191Type "show copying" to see the conditions.
2192There is absolutely no warranty for GDB.  Type "show warranty" for details.
2193This GDB was configured as "s390-ibm-linux"...
2194Core was generated by `./gdb'.
2195Program terminated with signal 11, Segmentation fault.
2196Reading symbols from /usr/lib/
2197Reading symbols from /lib/
2198Reading symbols from /lib/
2199Reading symbols from /lib/
2200#0  0x40126d1a in read () from /lib/
2201Setting up the environment for debugging gdb.
2202Breakpoint 1 at 0x4dc6f8: file utils.c, line 471.
2203Breakpoint 2 at 0x4d87a4: file top.c, line 2609.
2204(top-gdb) info stack
2205#0  0x40126d1a in read () from /lib/
2206#1  0x528f26 in rl_getc (stream=0x7ffffde8) at input.c:402
2207#2  0x528ed0 in rl_read_key () at input.c:381
2208#3  0x5167e6 in readline_internal_char () at readline.c:454
2209#4  0x5168ee in readline_internal_charloop () at readline.c:507
2210#5  0x51692c in readline_internal () at readline.c:521
2211#6  0x5164fe in readline (prompt=0x7ffff810 "\177x\177\177x")
2212    at readline.c:349
2213#7  0x4d7a8a in command_line_input (prompt=0x564420 "(gdb) ", repeat=1,
2214    annotation_suffix=0x4d6b44 "prompt") at top.c:2091
2215#8  0x4d6cf0 in command_loop () at top.c:1345
2216#9  0x4e25bc in main (argc=1, argv=0x7ffffdf4) at main.c:635
2221This is a program which lists the shared libraries which a library needs,
2222Note you also get the relocations of the shared library text segments which
2223help when using objdump --source.
2225 ldd ./gdb
2226outputs => /usr/lib/ (0x40018000) => /lib/ (0x4005e000) => /lib/ (0x40084000)
2230/lib/ => /lib/ (0x40000000)
2233Debugging shared libraries
2235Most programs use shared libraries, however it can be very painful
2236when you single step instruction into a function like printf for the 
2237first time & you end up in functions like _dl_runtime_resolve this is
2238the doing lazy binding, lazy binding is a concept in ELF where 
2239shared library functions are not loaded into memory unless they are 
2240actually used, great for saving memory but a pain to debug.
2241To get around this either relink the program -static or exit gdb type 
2242export LD_BIND_NOW=true this will stop lazy binding & restart the gdb'ing 
2243the program in question.
2247Debugging modules
2249As modules are dynamically loaded into the kernel their address can be
2250anywhere to get around this use the -m option with insmod to emit a load
2251map which can be piped into a file if required.
2253The proc file system
2255What is it ?.
2256It is a filesystem created by the kernel with files which are created on demand
2257by the kernel if read, or can be used to modify kernel parameters,
2258it is a powerful concept.
2262cat /proc/sys/net/ipv4/ip_forward 
2263On my machine outputs 
2265telling me ip_forwarding is not on to switch it on I can do
2266echo 1 >  /proc/sys/net/ipv4/ip_forward
2267cat it again
2268cat /proc/sys/net/ipv4/ip_forward 
2269On my machine now outputs
2271IP forwarding is on.
2272There is a lot of useful info in here best found by going in & having a look around,
2273so I'll take you through some entries I consider important.
2275All the processes running on the machine have their own entry defined by
2277So lets have a look at the init process
2278cd /proc/1
2280cat cmdline
2282init [2]
2284cd /proc/1/fd
2285This contains numerical entries of all the open files,
2286some of these you can cat e.g. stdout (2)
2288cat /proc/29/maps
2289on my machine emits
229100400000-00478000 r-xp 00000000 5f:00 4103       /bin/bash
229200478000-0047e000 rw-p 00077000 5f:00 4103       /bin/bash
22930047e000-00492000 rwxp 00000000 00:00 0
229440000000-40015000 r-xp 00000000 5f:00 14382      /lib/
229540015000-40016000 rw-p 00014000 5f:00 14382      /lib/
229640016000-40017000 rwxp 00000000 00:00 0
229740017000-40018000 rw-p 00000000 00:00 0
229840018000-4001b000 r-xp 00000000 5f:00 14435      /lib/
22994001b000-4001c000 rw-p 00002000 5f:00 14435      /lib/
23004001c000-4010d000 r-xp 00000000 5f:00 14387      /lib/
23014010d000-40111000 rw-p 000f0000 5f:00 14387      /lib/
230240111000-40114000 rw-p 00000000 00:00 0
230340114000-4011e000 r-xp 00000000 5f:00 14408      /lib/
23044011e000-4011f000 rw-p 00009000 5f:00 14408      /lib/
23057fffd000-80000000 rwxp ffffe000 00:00 0
2308Showing us the shared libraries init uses where they are in memory
2309& memory access permissions for each virtual memory area.
2311/proc/1/cwd is a softlink to the current working directory.
2312/proc/1/root is the root of the filesystem for this process. 
2314/proc/1/mem is the current running processes memory which you
2315can read & write to like a file.
2316strace uses this sometimes as it is a bit faster than the
2317rather inefficient ptrace interface for peeking at DATA.
2320cat status 
2322Name:   init
2323State:  S (sleeping)
2324Pid:    1
2325PPid:   0
2326Uid:    0       0       0       0
2327Gid:    0       0       0       0
2329VmSize:      408 kB
2330VmLck:         0 kB
2331VmRSS:       208 kB
2332VmData:       24 kB
2333VmStk:         8 kB
2334VmExe:       368 kB
2335VmLib:         0 kB
2336SigPnd: 0000000000000000
2337SigBlk: 0000000000000000
2338SigIgn: 7fffffffd7f0d8fc
2339SigCgt: 00000000280b2603
2340CapInh: 00000000fffffeff
2341CapPrm: 00000000ffffffff
2342CapEff: 00000000fffffeff
2344User PSW:    070de000 80414146
2345task: 004b6000 tss: 004b62d8 ksp: 004b7ca8 pt_regs: 004b7f68
2346User GPRS:
234700000400  00000000  0000000b  7ffffa90
234800000000  00000000  00000000  0045d9f4
23490045cafc  7ffffa90  7fffff18  0045cb08
235000010400  804039e8  80403af8  7ffff8b0
2351User ACRS:
235200000000  00000000  00000000  00000000
235300000001  00000000  00000000  00000000
235400000000  00000000  00000000  00000000
235500000000  00000000  00000000  00000000
2356Kernel BackChain  CallChain    BackChain  CallChain
2357       004b7ca8   8002bd0c     004b7d18   8002b92c
2358       004b7db8   8005cd50     004b7e38   8005d12a
2359       004b7f08   80019114                     
2360Showing among other things memory usage & status of some signals &
2361the processes'es registers from the kernel task_structure
2362as well as a backchain which may be useful if a process crashes
2363in the kernel for some unknown reason.
2365Some driver debugging techniques
2367debug feature
2369Some of our drivers now support a "debug feature" in
2370/proc/s390dbf see s390dbf.txt in the linux/Documentation directory
2371for more info.
2373to switch on the lcs "debug feature"
2374echo 5 > /proc/s390dbf/lcs/level
2375& then after the error occurred.
2376cat /proc/s390dbf/lcs/sprintf >/logfile
2377the logfile now contains some information which may help
2378tech support resolve a problem in the field.
2382high level debugging network drivers
2384ifconfig is a quite useful command
2385it gives the current state of network drivers.
2387If you suspect your network device driver is dead
2388one way to check is type 
2389ifconfig <network device> 
2390e.g. tr0
2391You should see something like
2392tr0       Link encap:16/4 Mbps Token Ring (New)  HWaddr 00:04:AC:20:8E:48
2393          inet addr:  Bcast:  Mask:
2394          UP BROADCAST RUNNING MULTICAST  MTU:2000  Metric:1
2395          RX packets:246134 errors:0 dropped:0 overruns:0 frame:0
2396          TX packets:5 errors:0 dropped:0 overruns:0 carrier:0
2397          collisions:0 txqueuelen:100
2399if the device doesn't say up
2401/etc/rc.d/init.d/network start 
2402( this starts the network stack & hopefully calls ifconfig tr0 up ).
2403ifconfig looks at the output of /proc/net/dev & presents it in a more presentable form
2404Now ping the device from a machine in the same subnet.
2405if the RX packets count & TX packets counts don't increment you probably
2406have problems.
2408cat /proc/net/arp
2409Do you see any hardware addresses in the cache if not you may have problems.
2410Next try
2411ping -c 5 <broadcast_addr> i.e. the Bcast field above in the output of
2412ifconfig. Do you see any replies from machines other than the local machine
2413if not you may have problems. also if the TX packets count in ifconfig
2414hasn't incremented either you have serious problems in your driver 
2415(e.g. the txbusy field of the network device being stuck on ) 
2416or you may have multiple network devices connected.
2421There is a new device layer for channel devices, some
2422drivers e.g. lcs are registered with this layer.
2423If the device uses the channel device layer you'll be
2424able to find what interrupts it uses & the current state 
2425of the device.
2426See the manpage chandev.8 &type cat /proc/chandev for more info.
2430Starting points for debugging scripting languages etc.
2435bash -x <scriptname>
2436e.g. bash -x /usr/bin/bashbug
2437displays the following lines as it executes them.
2438+ MACHINE=i586
2439+ OS=linux-gnu
2440+ CC=gcc
2441+ CFLAGS= -DPROGRAM='bash' -DHOSTTYPE='i586' -DOSTYPE='linux-gnu' -DMACHTYPE='i586-pc-linux-gnu' -DSHELL -DHAVE_CONFIG_H   -I. -I. -I./lib -O2 -pipe
2442+ RELEASE=2.01
2444+ RELSTATUS=release
2445+ MACHTYPE=i586-pc-linux-gnu   
2447perl -d <scriptname> runs the perlscript in a fully interactive debugger
2448<like gdb>.
2449Type 'h' in the debugger for help.
2451for debugging java type
2452jdb <filename> another fully interactive gdb style debugger.
2453& type ? in the debugger for help.
2459This is now supported by linux for s/390 & z/Architecture.
2460To enable it do compile the kernel with 
2461Kernel Hacking -> Magic SysRq Key Enabled
2462echo "1" > /proc/sys/kernel/sysrq
2463also type
2464echo "8" >/proc/sys/kernel/printk
2465To make printk output go to console.
2466On 390 all commands are prefixed with
2469^-t will show tasks.
2470^-? or some unknown command will display help.
2471The sysrq key reading is very picky ( I have to type the keys in an
2472 xterm session & paste them  into the x3270 console )
2473& it may be wise to predefine the keys as described in the VM hints above
2475This is particularly useful for syncing disks unmounting & rebooting
2476if the machine gets partially hung.
2478Read Documentation/sysrq.txt for more info
2482Enterprise Systems Architecture Reference Summary
2483Enterprise Systems Architecture Principles of Operation
2484Hartmut Penners s390 stack frame sheet.
2485IBM Mainframe Channel Attachment a technology brief from a CISCO webpage
2486Various bits of man & info pages of Linux.
2487Linux & GDB source.
2488Various info & man pages.
2489CMS Help on tracing commands.
2490Linux for s/390 Elf Application Binary Interface
2491Linux for z/Series Elf Application Binary Interface ( Both Highly Recommended )
2492z/Architecture Principles of Operation SA22-7832-00
2493Enterprise Systems Architecture/390 Reference Summary SA22-7209-01 & the
2494Enterprise Systems Architecture/390 Principles of Operation SA22-7201-05
2496Special Thanks
2498Special thanks to Neale Ferguson who maintains a much
2499prettier HTML version of this page at
2501Bob Grainger Stefan Bader & others for reporting bugs
2502 kindly hosted by Redpill Linpro AS, provider of Linux consulting and operations services since 1995.