stress-ng: update to 0.14.03 #19048
stress-ng: update to 0.14.03 #19048
Conversation
|
is this build error new? https://github.com/openwrt/packages/runs/7495316112?check_suite_focus=true i am not sure if this was present; i'll need to see if we can somehow fix it (if it wasn't broken before); |
|
other than ARC, rest is LGTM |
|
It's new, the old version succeeded: https://github.com/openwrt/packages/runs/4019861942?check_suite_focus=true Odd that this fails on ARC, the relevant Linux headers seem to be arch-independent... |
|
Probably a result of glibc update. |
|
do you mind pulling this patch into this PR? apologies for taking too long to get back, but this did require a bit of in-depth investigation to get to this point, even though the fix was relatively obvious; tbh: i have no idea why this was working before; but it does look like __NR_fstat64 can be defined while __NR_fstat may be undefined; will also see if the maintainer agrees with me thanks |
guijan
force-pushed
the
update-stress_ng
branch
from
6bcab25 to
b8d375c
Compare
|
Sorry, I didn't test adding the patch, I'll do it properly. |
Signed-off-by: Guilherme Janczak <[email protected]> Co-authored-by: Alexandru Ardelean <[email protected]>
guijan
force-pushed
the
update-stress_ng
branch
from
b8d375c to
9a85437
Compare
It seems that __NR_fstat64 may be defined while __NR_fstat may be undefined. Looking at the kernel file 'include/uapi/asm-generic/unistd.h' this looks possible where there is a 32-bit arch. In any case, checking that __NR_fstat64 exists/has-a-value when adding it to the table looks like a good idea, vs adding __NR_fstat. This was found when updating stress-ng here: openwrt/packages#19048 The ARCHS (or ARC) build is failing. No idea why it's the only one. But this patch fixes it. Signed-off-by: Alexandru Ardelean <[email protected]>
| -#if !defined(__PCC__) && \ | ||
| - !defined(__TINYC__) | ||
| -#define _FORTIFY_SOURCE 2 | ||
| -#endif |
There was a problem hiding this comment.
this patch is fine, but i don't see it fixing errors;
it's fixing warnings;
if possible i would send it upstream to have it discussed with them;
they are reasonably quick;
i would not carry patches in our tree if they don't fix anything more grave;
Signed-off-by: Guilherme Janczak [email protected]
Maintainer: @commodo
Compile tested: x64, latest snapshot SDK as of today