Fwd: How to debug on leon3 with sis and gdb

Fri May 21 08:35:08 UTC 2021

Forgot copy the list.

On 5/21/21 7:02 AM, Richi Dubey wrote:
> Hi,
>
> A test fails at the assert(sc==RTEMS_SUCCESSFUL) after a call to rtems_task_set_affinity.
>
> So, following the docs <https://devel.rtems.org/wiki/Debugging/sis>, I figured out where the break point should be set:
>
> ---------------------------------------------------
>
> $ ~/quick-start/rtems/6/bin/sparc-rtems6-sis -leon3 -m 4 ~/sis-quick-start/src/rtems/build/sparc/leon3/testsuites/smptests/smpschedstrongapa01.exe
>
>  SIS - SPARC/RISCV instruction simulator 2.26,  copyright Jiri Gaisler 2020
>  Bug-reports to jiri at gaisler.se <mailto:jiri at gaisler.se>
>
>  LEON3 emulation enabled, 4 cpus online, delta 50 clocks
>
>  Loaded /home/richi/sis-quick-start/src/rtems/build/sparc/leon3/testsuites/smptests/smpschedstrongapa01.exe, entry 0x40000000
> cpu0> hi 10
> trace history length = 10
> cpu0> run
> Waking CPU 1
> Waking CPU 2
> Waking CPU 3
> assertion "sc == RTEMS_SUCCESSFUL" failed: file "../../../testsuites/smptests/smpschedstrongapa01/init.c", line 264, function: Init
> cpu 0 in error mode (tt = 0x80)
>    423950  4000f9e0:  91d02000   ta  0x0
> cpu0> hi
>    423922  40019d40:  90100018   mov  %i0, %o0
>    423923  4000f9e4:  82102001   mov  1, %g1
>    423924  4000f9e8:  84100008   mov  %o0, %g2
>    423925  4000f9ec:  86100009   mov  %o1, %g3
>    423926  4000f9f0:  91d02000   ta  0x0
>    423928  40000800:  a1480000   mov  %psr, %l0
>    423929  40000804:  2910003e   sethi  %hi(0x4000f800), %l4
>    423930  40000808:  81c521e0   jmp  %l4 + 0x1e0
>    423932  4000080c:  a6102080   mov  128, %l3
>    423933  4000f9e0:  91d02000   ta  0x0
> cpu0> reg
>
>  INS       LOCALS      OUTS     GLOBALS
>    0:  00000007   F3001FC6   00000007   00000000
>    1:  4002F008   4000F9F0   00000000   00000001
>    2:  40029CF8   4000F9F4   4002F008   00000007
>    3:  0000000A   00000080   40020400   4002F008
>    4:  00000073   4000F800   0000000E   00000000
>    5:  40029C00   00000000   00000004   00000000
>    6:  4002EEE0   4002A250   4002EE78   4002CD00
>    7:  40019D3C   4002CD00   4000EC80   00000000
>
>  psr: F3001FC6   wim: 00000008   tbr: 40000800   y: 00000000
>
>   pc: 4000F9E0 = 91D02000     ta  0x0
>  npc: 4000F9E4 = 82102001     mov  1, %g1
>  IU in error mode
>
> cpu0> quit
> ---------------------------------------------------
> So the breakpoint has to be set at the second last instruction at address 0x4000080c

This is the wrong assumption. The program does not halt where the assertion fails, it prints the error message and exits the normal way. To stop execution right where the assertion is made, you should add a breakpoint at the line number:

bre init.c:264

Also, make sure you compile RTEMS without optimization, or single-stepping and debugging will be hard. I usually do this by removing all optimization flags in the OPTIMIZATION_FLAGS variable in the bsp .ini file before configuring with waf.

>
> But when I try to debug this with gdb keeping the sis as remote target, it does not work:
>
> -------------------------
> $ ~/quick-start/rtems/6/bin/sparc-rtems6-gdb ~/sis-quick-start/src/rtems/build/sparc/leon3/testsuites/smptests/smpschedstrongapa01.exe
> GNU gdb (GDB) 10.1.90.20210409-git
> Copyright (C) 2021 Free Software Foundation, Inc.
> License GPLv3+: GNU GPL version 3 or later <http://gnu.org/licenses/gpl.html <http://gnu.org/licenses/gpl.html>>
> This is free software: you are free to change and redistribute it.
> There is NO WARRANTY, to the extent permitted by law.
> Type "show copying" and "show warranty" for details.
> This GDB was configured as "--host=x86_64-linux-gnu --target=sparc-rtems6".
> Type "show configuration" for configuration details.
> For bug reporting instructions, please see:
> <https://www.gnu.org/software/gdb/bugs/ <https://www.gnu.org/software/gdb/bugs/>>.
> Find the GDB manual and other documentation resources online at:
>     <http://www.gnu.org/software/gdb/documentation/ <http://www.gnu.org/software/gdb/documentation/>>.
>
> For help, type "help".
> Type "apropos word" to search for commands related to "word"...
> Reading symbols from /home/richi/sis-quick-start/src/rtems/build/sparc/leon3/testsuites/smptests/smpschedstrongapa01.exe...
> (gdb) tar sim -leon3
> Undefined target command: "sim -leon3".  Try "help target".
> (gdb) target extended-remote localhost:1234
> Remote debugging using localhost:1234
> 0x00000000 in ?? ()
> (gdb) load
> Loading section .text, size 0x218f0 lma 0x40000000
> Loading section .rtemsroset, size 0x90 lma 0x400218f0
> Loading section .data, size 0x530 lma 0x40029980
> Start address 0x40000000, load size 138928
> Transfer rate: 3083 KB/sec, 271 bytes/write.
> (gdb) bre 0x4000080c

To break at a hex address in gdb, you need a pointer:

bre *0x4000080c

> Function "0x4000080c" not defined.
> Make breakpoint pending on future shared library load? (y or [n]) y
> Breakpoint 1 (0x4000080c) pending.
> (gdb) run
> The program being debugged has been started already.
> Start it from the beginning? (y or n) y
> Starting program: /home/richi/sis-quick-start/src/rtems/build/sparc/leon3/testsuites/smptests/smpschedstrongapa01.exe
>
> Program received signal SIGTERM, Terminated.
> syscall () at ../../../cpukit/score/cpu/sparc/syscall.S:44
> 44 ta 0 ! syscall 1, halt with %g1,%g2,%g3 info
> (gdb)
>
> --------------------
>
> And the output on the other terminal is:
>
> ------------------------------------
> $ ~/quick-start/rtems/6/bin/sparc-rtems6-sis -leon3 -m 4 -gdb
>
>  SIS - SPARC/RISCV instruction simulator 2.26,  copyright Jiri Gaisler 2020
>  Bug-reports to jiri at gaisler.se <mailto:jiri at gaisler.se>
>
>  LEON3 emulation enabled, 4 cpus online, delta 50 clocks
>
> gdb: listening on port 1234 connected
> X40000000,0:#72
> Waking CPU 1
> Waking CPU 2
> Waking CPU 3
> assertion "sc == RTEMS_SUCCESSFUL" failed: file "../../../testsuites/smptests/smpschedstrongapa01/init.c", line 264, function: Init
> ------------------------------------
>
> So, is there a way I can debug this? Can I use si or ni, cause it is not working here when I tried it after 'load'. Also can I not set breakpoints at functions and then step continue till the function comes?
>
> Please advise.
> Thanks.
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.rtems.org/pipermail/devel/attachments/20210521/0f0e49f7/attachment.html>