Maybe I was finally successful:
stadler seems to have come into troubles after runing a single (faulty) executable
the executable is in
/home/pierre/pas/test directory
its name is
./tweaklib2
it uses a (probably faulty) library called
./libtweaklib1.so
both are 64-bit executable generated with the current Free Pascal compiler
on my account, located in directory
/home/pierre/pas/fpc-3.1.1/bin/ppcsparc64 (cross compiler from 32 to 64 bit)
Last things I did was
recompiling those two sources, using
pierre@stadler:~/pas/test$ ppcsparc64 -gl tweaklib1.pp -k-rpath -k. -Cg -n -Fu../trunk/fpcsrc/rtl/units/sparc64-linux
pierre@stadler:~/pas/test$ ppcsparc64 -gl tweaklib2.pp -k-rpath -k. -Cg -n -Fu../trunk/fpcsrc/rtl/units/sparc64-linux
Running simply ./tweaklib2
generated a SIGILL which stopped the program without any core generation...
Running it inside gdb did not help,
SIGILL was not caught by GDB :-(
Running it with strace generated this message:
rt_sigaction(SIGBUS, {sa_handler=SIG_DFL, sa_mask=[], sa_flags=0}, NULL, 0xffff800100134ca8, 8) = 0
rt_sigaction(SIGILL, {sa_handler=SIG_DFL, sa_mask=[], sa_flags=0}, NULL, 0xffff800100134ca8, 8) = 0
Message from syslogd@stadler at Jul 18 20:04:02 ...
kernel:[206085.986213] usercopy: Kernel memory exposure attempt detected from process stack (offset 0, size 128)!
I tried to access via a second ssh session to the machine,
but the session froze also, and I finally got completely kicked out,
probably due to a system shutdown...
This kernel crash might be easier to reproduce than any of what I got before ...
I am still unable to reconnect to the machine,
so maybe someone can resart the machine,
and test if running the tweaklib2 in /home/pierre/pas/test directory
several times or with strace is sufficient to get a 'reproducible' crash.
In the hope that this will help,
Pierre Muller
Le 10/07/2018 à 09:52, John Paul Adrian Glaubitz a écrit :
> Hi!
>
> FreePascal is using one of Debian's sparc64 machines to do CI for
> the development of their compiler FPC.
>
> The testsuite of FPC has been known to cause kernel issues and even
> crash the machine when running on Debian sparc64. While the number of
> crashes has been reduced in the past, we're still seeing some kernel
> issues from time to time:
>
> [285818.656472] usercopy: Kernel memory exposure attempt detected from null address (offset 0, size 128)!
> [285818.656650] kernel BUG at /build/linux-UzksCq/linux-4.17.3/mm/usercopy.c:100!
> [285818.656662] \|/ ____ \|/
> "@'/ .. \`@"
> /_| \__/ |_\
> \__U_/
> [285818.656671] ld-linux.so.2(18698): Kernel bad sw trap 5 [#3]
> [285818.656689] CPU: 0 PID: 18698 Comm: ld-linux.so.2 Tainted: G D 4.17.0-1-sparc64-smp #1 Debian 4.17.3-1
> [285818.656702] TSTATE: 0000004411001602 TPC: 0000000000634500 TNPC: 0000000000634504 Y: 00000001 Tainted: G D
> [285818.656728] TPC: <usercopy_abort+0x80/0xa0>
> [285818.656738] g0: ffff80052a534000 g1: 0000000000000000 g2: 0000000000000007 g3: 0000000000000000
> [285818.656747] g4: ffff80050734f080 g5: ffff8007fe67e000 g6: ffff80052a534000 g7: 000000000000000e
> [285818.656756] o0: 0000000000b94780 o1: 0000000000000064 o2: 0000000000bd5d10 o3: 0000000000b94818
> [285818.656765] o4: 0000000000baac80 o5: 0000000000baac80 sp: ffff80052a537061 ret_pc: 00000000006344f8
> [285818.656778] RPC: <usercopy_abort+0x78/0xa0>
> [285818.656789] l0: ffff80052a534018 l1: 0000000000000003 l2: ffff80052a534000 l3: ffff80052a5343c8
> [285818.656799] l4: 0000000000000000 l5: 000007feffa82000 l6: ffff80052a534000 l7: 0000000011001001
> [285818.656808] i0: 0000000000b94818 i1: 0000000000baac80 i2: 0000000000000001 i3: 0000000000000000
> [285818.656818] i4: 0000000000000080 i5: 000007feffa83671 i6: ffff80052a537131 i7: 00000000006346e0
> [285818.656832] I7: <__check_object_size+0x1c0/0x220>
> [285818.656837] Call Trace:
> [285818.656852] [00000000006346e0] __check_object_size+0x1c0/0x220
> [285818.656874] [000000000042dfb0] synchronize_user_stack+0xb0/0x180
> [285818.656889] [000000000042dfc4] synchronize_user_stack+0xc4/0x180
> [285818.656904] [000000000042dfc4] synchronize_user_stack+0xc4/0x180
> [285818.656918] [000000000042dfc4] synchronize_user_stack+0xc4/0x180
> [285818.656934] [000000000042dfc4] synchronize_user_stack+0xc4/0x180
> [285818.656949] [000000000042e600] do_signal+0x60/0x480
> [285818.656963] [000000000042f270] do_notify_resume+0x50/0xa0
> [285818.656978] [0000000000404b44] __handle_signal+0xc/0x2c
> [285818.656996] Caller[00000000006346e0]: __check_object_size+0x1c0/0x220
> [285818.657013] Caller[000000000042dfb0]: synchronize_user_stack+0xb0/0x180
> [285818.657028] Caller[000000000042dfc4]: synchronize_user_stack+0xc4/0x180
> [285818.657042] Caller[000000000042dfc4]: synchronize_user_stack+0xc4/0x180
> [285818.657056] Caller[000000000042dfc4]: synchronize_user_stack+0xc4/0x180
> [285818.657071] Caller[000000000042dfc4]: synchronize_user_stack+0xc4/0x180
> [285818.657085] Caller[000000000042e600]: do_signal+0x60/0x480
> [285818.657099] Caller[000000000042f270]: do_notify_resume+0x50/0xa0
> [285818.657111] Caller[0000000000404b44]: __handle_signal+0xc/0x2c
> [285818.657123] Caller[ffff8001001296dc]: 0xffff8001001296dc
> [285818.657128] Instruction DUMP:
> [285818.657134] 92102064
> [285818.657140] 7ff7d142
> [285818.657145] 90122380
> [285818.657151] <91d02005>
> [285818.657157] 33002eab
> [285818.657162] b2166080
> [285818.657168] 98100019
> [285818.657173] 106ffff0
> [285818.657179] 82100019
>
> [309379.910290] Unable to handle kernel NULL pointer dereference
> [309379.910457] tsk->{mm,active_mm}->context = 00000000000009d2
> [309379.910554] tsk->{mm,active_mm}->pgd = ffff80067e2e4000
> [309379.910601] \|/ ____ \|/
> "@'/ .. \`@"
> /_| \__/ |_\
> \__U_/
> [309379.910610] ppcsparc(8458): Oops [#4]
> [309379.910628] CPU: 10 PID: 8458 Comm: ppcsparc Tainted: G D 4.17.0-1-sparc64-smp #1 Debian 4.17.3-1
> [309379.910640] TSTATE: 0000004423001603 TPC: 0000000000a6cccc TNPC: 0000000000a6ccd0 Y: 00000000 Tainted: G D
> [309379.910659] TPC: <NGcopy_to_user+0x28c/0x4c0>
> [309379.910668] g0: 000000000044d044 g1: 0000000000000080 g2: 0000000000000020 g3: 0000000000000030
> [309379.910677] g4: ffff8007e10aa680 g5: ffff8007fe7be000 g6: ffff80052a3ec000 g7: fffffffffffffff2
> [309379.910686] o0: ffff80052a3ec500 o1: 0000000000000040 o2: 0000000000000050 o3: 00000000ffaabfd0
> [309379.910694] o4: 000000000000002f o5: fffffffffffffff2 sp: ffff80052a3ef171 ret_pc: 0000000000000010
> [309379.910703] RPC: <0x10>
> [309379.910813] l0: 00000000ffaabfd0 l1: 00000000ffaabf68 l2: 00000000004076ac l3: 0000000000000000
> [309379.910847] l4: 0000000000000000 l5: 00000000f4092000 l6: ffff80052a3ec000 l7: 0000000011001005
> [309379.910866] i0: ffff80052a3ec500 i1: 0000000000000550 i2: 0000000000000000 i3: ffff80052a3ec550
> [309379.910885] i4: 0000000000000000 i5: 0000000000000005 i6: ffff80052a3ef1f1 i7: 000000000042f304
> [309379.910928] I7: <save_fpu_state+0x44/0xa0>
> [309379.910944] Call Trace:
> [309379.910975] [000000000042f304] save_fpu_state+0x44/0xa0
> [309379.911000] [000000000042f2e8] save_fpu_state+0x28/0xa0
> [309379.911031] [000000000044d640] do_signal32+0x880/0x980
> [309379.911055] [000000000042e714] do_signal+0x174/0x480
> [309379.911078] [000000000042f270] do_notify_resume+0x50/0xa0
> [309379.911097] [0000000000404b44] __handle_signal+0xc/0x2c
> [309379.911113] Caller[000000000042f304]: save_fpu_state+0x44/0xa0
> [309379.911125] Caller[000000000042f2e8]: save_fpu_state+0x28/0xa0
> [309379.911142] Caller[000000000044d640]: do_signal32+0x880/0x980
> [309379.911156] Caller[000000000042e714]: do_signal+0x174/0x480
> [309379.911169] Caller[000000000042f270]: do_notify_resume+0x50/0xa0
> [309379.911180] Caller[0000000000404b44]: __handle_signal+0xc/0x2c
> [309379.911189] Caller[0000000000030730]: 0x30730
> [309379.911194] Instruction DUMP:
> [309379.911199] 84102020
> [309379.911205] 86102030
> [309379.911210] 92102040
> [309379.911215] <d89e5c40>
> [309379.911220] d49e5c4f
> [309379.911226] c36e4009
> [309379.911232] d8f22000
> [309379.911238] daf22008
> [309379.911244] d89e5c42
>
> Does anyone have a suggestion on how to debug this?
>
> Thanks,
> Adrian
>
--
To unsubscribe from this list: send the line "unsubscribe sparclinux" in
the body of a message to majordomo@xxxxxxxxxxxxxxx
More majordomo info at http://vger.kernel.org/majordomo-info.html
