ldv changed the topic of #strace to: https://strace.io | https://strace.io/logs/ | strace 5.2 is out | strace-devel@lists.strace.io for dev discussions
<mator> ldv, thanks
<mator> ldv, getting the following while compiling kernel:
<mator> <stdin>:1511:2: warning: #warning syscall clone3 not implemented [-Wcpp]
<mator> trying to bisect kernel which git commit introduced this warning
<ldv> I bet it's "fork: add clone3" merged after v5.2
<ldv> But the test is not about clone3
<ldv> But the most likely cause of this kernel regression on sparc64 is the addition of clone3
<mator> yep, so i went forward...
<mator> Bisecting: 20 revisions left to test after this (roughly 4 steps)
<mator> as well nft goes 'sigbus' in libnftnl
<mator> off to shopping mall... need to buy groceries and eat something
<mator> b3e5838252665ee4cfa76b82bdf1198dca81e5be is the first bad commit
<ldv> mator: I don't think so
<ldv> can you check whether commit 9014143bab2f3bc0b9e5db3bc8d00e2a43e50fbd works?
<mator> sure, wait for 15-20 minutes till kernel compiles...
<ldv> alternatively, you can change "ptid = -1;" to "ptid = 0;" in my test
<ldv> with this change the test used to work on at least one of your kernels containing b3e5838252665ee4cfa76b82bdf1198dca81e5be
<mator> ldv, works on 9014143bab2f3bc0b9e5db3bc8d00e2a43e50fbd
<mator> (no failure)
<mator> so i've no idea how to debug this issue
<ldv> my guess is that v5.2 also works
<ldv> also my guess is that the regression is somehow caused by the merge commit v5.2-3315-g8f6ccf6159aed1f04c6d179f61f6fb2691261e84 that introduced clone3
<mator> i could try to bisect with good v5.2-rc6-1-g9014143bab2f and master as bad
<mator> but that would take some time
<ldv> Have you built v5.2 already?
<ldv> 9014143bab2f3bc0b9e5db3bc8d00e2a43e50fbd is v5.2-rc7~19^2~3 so I'm pretty sure v5.2 also works
<mator> yep
<mator> Linux ttip 5.2.0-03311-g5450e8a316a6 #1019 SMP Thu Jul 11 12:12:18 MSK 2019 sparc64 GNU/Linux
<mator> CLONE_PIDFD: pid=1207 pidfd=3
<mator> mator@ttip:~$ uname -a
<mator> CLONE_PARENT_SETTID: pid=1206 ptid=1206
<mator> mator@ttip:~$ ./a.out
<ldv> yes, I expected this
<ldv> so it is 03315 then
<mator> what is the git hash of 03315 ?
<ldv> 8f6ccf6159aed1f04c6d179f61f6fb2691261e84
<ldv> the next merge commit after 03311
<mator> let me boot to 03315
<ldv> that merge commit contains non-trivial conflict resolution
<ldv> sure
<mator> current v5.2-3315-g8f6ccf6159ae build shows warning right after a few lines of a kernel compile log:
<mator> <stdin>:1511:2: warning: #warning syscall clone3 not implemented [-Wcpp]
<mator> CALL scripts/checksyscalls.sh
<mator> UPD include/generated/asm-offsets.h
<ldv> yes, this is expected, clone3 has not been wired up on sparc64 yet
<ldv> i hope it will be wired up on all relevant architectures before the next kernel is released
<mator> Linux ttip 5.2.0-03315-g8f6ccf6159ae #1066 SMP Sat Jul 13 22:04:34 MSK 2019 sparc64 GNU/Linux
<mator> a.out: clone CLONE_PIDFD|SIGCHLD: Bad address
<ldv> i don't know whether it makes sense to bisect two subsequent merge commits
<mator> up to you
<ldv> Christian would be thankful I think :)
<mator> so which one to check ?
<ldv> 5.2.0-03311-g5450e8a316a6 is good, 5.2.0-03315-g8f6ccf6159ae is bad
pombreda has quit [Ping timeout: 248 seconds]
pombreda has joined #strace
pombreda has quit [Ping timeout: 246 seconds]
<mator> ldv, i hope you will tell Christian, not me :)