Hello Ryutaroh Matsumoto, dear Maintainer, I am not involved in packaging valgrind, just trying to help with some random bug reports. For this report #983377 I cannot follow, how #928224 is blocking it?
#928224 is about valgrind not running at all, with "a function redirection ... cannot be set up" and exits. This was not reproducible for me in 2019 inside a VM. #983377 has "disInstr(thumb): unhandled instruction: 0xEC51 0x0F1E", which seems to be tracked upstream in [1]. A valgrind package build with the patch [2] from this bug does not show this "unhandled instruction" message anymore. The instruction seems to be on purpose by openssl [3] to detect certain cpu features. [1] https://bugs.kde.org/show_bug.cgi?id=344802 [2] https://bugs.kde.org/attachment.cgi?id=113735 [3] https://sources.debian.org/src/openssl/1.1.1j-1/crypto/armcap.c/#L224 Program received signal SIGILL, Illegal instruction. _armv7_tick () at crypto/armv4cpuid.S:136 136 crypto/armv4cpuid.S: Datei oder Verzeichnis nicht gefunden. 1: x/i $pc => 0x4b9f926 <_armv7_tick>: mrrc 15, 1, r0, r1, cr14 (gdb) bt #0 _armv7_tick () at crypto/armv4cpuid.S:136 #1 0x04b9a0fa in OPENSSL_cpuid_setup () at ../crypto/armcap.c:224 #2 0x0400bf40 in call_init (l=<optimized out>, argc=argc@entry=6, argv=argv@entry=0xbdf1c624, env=env@entry=0xbdf1c640) at dl-init.c:72 #3 0x0400bfe2 in call_init (env=<optimized out>, argv=<optimized out>, argc=<optimized out>, l=<optimized out>) at dl-init.c:30 #4 _dl_init (main_map=0x402a9b8, argc=6, argv=0xbdf1c624, env=0xbdf1c640) at dl-init.c:119 #5 0x04000ac4 in _dl_start_user () from /lib/ld-linux-armhf.so.3 #983377 shows also multiple "Invalid write ... below stack pointer". Might this be kind of a stack protection? There is an upstream bug describing something similar in Android and there was a way to silence such access integrated: '--ignore-range-below-sp=1152-0'. Maybe related, in [5] it looks like gcc was starting to emit stack protection instructions in 2017. [4] https://bugs.kde.org/show_bug.cgi?id=360571 [5] https://bugzilla.redhat.com/show_bug.cgi?id=1522678 Kind regards, Bernhard valgrind --vgdb=yes --vgdb-error=0 /home/benutzer/autopkgtest.7hddQs/autopkgtest_tmp/mock-sshd --user=joe --password=foobar --port 1234 -v gdb -q /home/benutzer/autopkgtest.7hddQs/autopkgtest_tmp/mock-sshd target remote | vgdb ==30911== Invalid write of size 4 ==30911== at 0x4964BF6: libssh_constructor (init.c:111) ==30911== by 0x400BF3F: call_init.part.0 (dl-init.c:72) ==30911== by 0x400BFE1: call_init (dl-init.c:30) ==30911== by 0x400BFE1: _dl_init (dl-init.c:119) ==30911== by 0x4000AC3: ??? (in /usr/lib/arm-linux-gnueabihf/ld-2.31.so) ==30911== Address 0xbda4d650 is on thread 1's stack ==30911== 8 bytes below stack pointer ==30911== ==30911== (action on error) vgdb me ... Program received signal SIGTRAP, Trace/breakpoint trap. 0x04964bf6 in libssh_constructor () at ./src/init.c:111 111 { 1: x/i $pc => 0x4964bf6 <libssh_constructor+10>: str.w r0, [r12, #4088] ; 0xff8 (gdb) disassemble libssh_constructor,libssh_constructor+40 Dump of assembler code from 0x4964bec to 0x4964c14: 0x04964bec <libssh_constructor+0>: push {r4, lr} 0x04964bee <libssh_constructor+2>: mov.w r12, #4096 ; 0x1000 0x04964bf2 <libssh_constructor+6>: sub.w r12, sp, r12 => 0x04964bf6 <libssh_constructor+10>: str.w r0, [r12, #4088] ; 0xff8 0x04964bfa <libssh_constructor+14>: ldr r2, [pc, #84] ; (0x4964c50 <libssh_constructor+100>) 0x04964bfc <libssh_constructor+16>: ldr r4, [pc, #84] ; (0x4964c54 <libssh_constructor+104>) 0x04964bfe <libssh_constructor+18>: add r2, pc End of assembler dump. (gdb) print/x $r0 $1 = 0x6 (gdb) print/x $r12 $2 = 0xbda4c658 (gdb) print/x $sp $3 = 0xbda4d658 (gdb) print/x $pc $4 = 0x4964bf7 (gdb) bt #0 0x04964bf6 in libssh_constructor () at ./src/init.c:111 #1 0x0400bf40 in call_init (l=<optimized out>, argc=argc@entry=6, argv=argv@entry=0xbda4d6a4, env=env@entry=0xbda4d6c0) at dl-init.c:72 #2 0x0400bfe2 in call_init (env=<optimized out>, argv=<optimized out>, argc=<optimized out>, l=<optimized out>) at dl-init.c:30 #3 _dl_init (main_map=0x402a9b8, argc=6, argv=0xbda4d6a4, env=0xbda4d6c0) at dl-init.c:119 #4 0x04000ac4 in _dl_start_user () from /lib/ld-linux-armhf.so.3 Backtrace stopped: previous frame identical to this frame (corrupt stack?)
