Hi Arnd,

On Mon, 25 Jan 2021 at 20:41, Arnd Bergmann [off-list ref] wrote:

On Mon, Jan 25, 2021 at 3:48 PM Naresh Kamboju
[off-list ref] wrote:

quoted

LTP syscalls madvise08 test case failed on all devices from
Linux next 20210118 to till day.
strace log attached to this email and link provided below.

BAD: next-20210118
GOOD: next-20210115

This failure is easily reproducible on Linux next tag 20210118 above.

tst_test.c:1250: TINFO: Timeout per run is 0h 15m 00s
madvise08.c:73: TINFO: Temporary core pattern is
'/scratch/ltp-2nftQzNI1K/HclFMH/dump-%p'
madvise08.c:112: TINFO: Dump file should be dump-10109
madvise08.c:196: TPASS: madvise(..., MADV_DONTDUMP)
madvise08.c:112: TINFO: Dump file should be dump-10110
madvise08.c:203: TFAIL: No sequence in dump after MADV_DODUMP.

strace log,
https://lkft.validation.linaro.org/scheduler/job/2184866#L1257

Ok, so in this part of the log,

[pid   485] --- SIGCHLD {si_signo=SIGCHLD, si_code=CLD_DUMPED,
si_pid=487, si_uid=0, si_status=SIGABRT, si_utime=0, si_stime=0} ---
[pid   485] write(2, \"madvise08.c:117: \33[1;34mTINFO: \33\"...,
64madvise08.c:117: [1;34mTINFO: [0mDump file should be dump-487
) = 64
[pid   485] access(\"dump-487\", F_OK)    = 0
[pid   485] openat(AT_FDCWD, \"dump-487\", O_RDONLY) = 3
[pid   485] read(3,
\"\177ELF\2\1\1\0\0\0\0\0\0\0\0\0\4\0>\0\1\0\0\0\0\0\0\0\0\0\0\0\"...,
1024) = 1024
[pid   485] read(3,
\"\0\320\3\0\0\0\0\0\0\340\375\24\304\177\0\0\0\0\0\0\0\0\0\0\0\20\0\0\0\0\0\0\"...,
1024) = 292
[pid   485] read(3, \"\", 1024)           = 0
[pid   485] close(3)                    = 0
[pid   485] write(2, \"madvise08.c:208: \33[1;31mTFAIL: \33\"...,
74madvise08.c:208: [1;31mTFAIL: [0mNo sequence in dump after
MADV_DODUMP.

it seems that the data that was requested to be dumped with MADV_DODUMP is
indeed completely absent.

There was exactly one commit that got merged between next-20210115 and
next-20120118
related to core dumps: 8a3cc755b138 ("coredump: don't bother with
do_truncate()").
Adding Al Viro to Cc for that.

Naresh, could you try reverting that patch?

This suspecting commit reverted and tested and the test case PASS.

commit 8a3cc755b13808eba74846dfd1033fcbc21f9a65
Author: Al Viro [off-list ref]
Date:   Sun Mar 8 09:16:37 2020 -0400

    coredump: don't bother with do_truncate()

    have dump_skip() just remember how much needs to be skipped,
    leave actual seeks/writing zeroes to the next dump_emit()
    or the end of coredump output, whichever comes first.
    And instead of playing with do_truncate() in the end, just
    write one NUL at the end of the last gap (if any).

    Signed-off-by: Al Viro [off-list ref]

 fs/binfmt_elf.c          |  1 -
 fs/coredump.c            | 56 +++++++++++++++++++++++++++---------------------
 include/linux/binfmts.h  |  1 +
 include/linux/coredump.h |  1 -


Test case output link,
https://lkft.validation.linaro.org/scheduler/job/2184975#L1369
https://lkft.validation.linaro.org/scheduler/job/2184972#L1358

- Naresh