[LTP] Linux-4.4: pwritev201.c:62: FAIL: pwritev2() failed: EINVAL

[Xiao Yang]

Hi Naresh,


Thank you for reporting the issue.

 From kernel code and manpage, pwritev2() has not been introduced on 
stable kernel v4.4.

glibc also provides pwritev2()Â  since glibc-2.26 so that you can still 
call pwritev2() by glibc.

But glibc doesn't handle 'offset == -1' correctly before glibc-2.28 
because it just replaces

pwritev2() with pwritev() and calling pwritev() with negative offset 
returns EINVAL.

See the following code in glibc:

----------------------------------------------------------------------------------------------------------------------- 


ssize_t
pwritev2 (int fd, const struct iovec *vector, int count, off_t offset,
 Â Â Â Â Â Â Â Â Â  int flags)
{
# ifdef __NR_pwritev2
 Â  ssize_t result = SYSCALL_CANCEL (pwritev2, fd, vector, count,
 Â Â Â Â Â Â Â Â Â Â Â Â Â Â Â Â Â Â Â Â Â Â Â Â Â Â Â Â Â Â Â Â Â Â  LO_HI_LONG (offset), flags);
 Â  if (result >= 0)
 Â Â Â  return result;
# endif
 Â  /* Trying to emulate the pwritev2 syscall flags is troublesome:

 Â Â Â Â  * We can not temporary change the file state of the O_DSYNC and 
O_SYNC
 Â Â Â Â Â Â  flags to emulate RWF_{D}SYNC (attempts to change the state of using
 Â Â Â Â Â Â  fcntl are silently ignored).

 Â Â Â Â  * IOCB_HIPRI requires the file opened in O_DIRECT and uses an 
internal
 Â Â Â Â Â Â  semantic not provided by any other flag (O_NONBLOCK for 
instance).  */

 Â  if (flags != 0)
 Â Â Â  {
 Â Â Â Â Â  __set_errno (ENOTSUP);
 Â Â Â Â Â  return -1;
 Â Â Â  }
 Â  return pwritev (fd, vector, count, offset);
}
----------------------------------------------------------------------------------------------------------------------- 


This issue has been fixed by commit d4b4a00 since glibc-2.28, so I think 
you should use

glibc-2.28 or newer to test pwritev201 on stable kernel v4.4. Could you 
test pwritev201

with newer glibc again on stable kernel v4.4?

I think it is a known bug of glibc rather than a bug of test case.


Best Regards,

Xiao Yang

On 06/17/2019 10:30 AM, Naresh Kamboju wrote:
> Newly added tests pwritev201 and pwritev201_64 from LTP version
> 20190517 got failed on all devices for Linux 4.4
> do you test this on linux stable  4.4 branch ? do you notice this failure ?
>
> Test PASS on stable 4.9, 4.14 and 4.19 branches.
>
> tst_test.c:1096: INFO: Timeout per run is 0h 15m 00s
> pwritev201.c:96: PASS: pwritev2() wrote 64 bytes successfully with
> content 'a' expectedly
> pwritev201.c:96: PASS: pwritev2() wrote 64 bytes successfully with
> content 'a' expectedly
> pwritev201.c:96: PASS: pwritev2() wrote 64 bytes successfully with
> content 'a' expectedly
> pwritev201.c:62: FAIL: pwritev2() failed: EINVAL
> pwritev201.c:62: FAIL: pwritev2() failed: EINVAL
> pwritev201.c:62: FAIL: pwritev2() failed: EINVAL
> Summary:
> passed   3
> failed   3
> skipped  0
> warnings 0
>
> tst_test.c:1096: INFO: Timeout per run is 0h 15m 00s
> pwritev201.c:96: PASS: pwritev2() wrote 64 bytes successfully with
> content 'a' expectedly
> pwritev201.c:96: PASS: pwritev2() wrote 64 bytes successfully with
> content 'a' expectedly
> pwritev201.c:96: PASS: pwritev2() wrote 64 bytes successfully with
> content 'a' expectedly
> pwritev201.c:62: FAIL: pwritev2() failed: EINVAL
> pwritev201.c:62: FAIL: pwritev2() failed: EINVAL
> pwritev201.c:62: FAIL: pwritev2() failed: EINVAL
> Summary:
> passed   3
> failed   3
> skipped  0
> warnings 0
>
> Full test log,
> https://lkft.validation.linaro.org/scheduler/job/751027#L9608
>
> Test results comparison link,
> https://qa-reports.linaro.org/_/comparetest/?project=22&project=6&project=58&project=135&project=40&project=23&project=167&suite=ltp-syscalls-tests&test=pwritev201
>
> Best regards
> Naresh Kamboju
>