Comments (11)
ColinIanKing
commented on June 2, 2024
1
Fixed in commit d533345 - please re-try with this fix and let me know if it helps.
from stress-ng.
ColinIanKing
commented on June 2, 2024
1
Order fixed in commit 20b483e
from stress-ng.
vt-alt
commented on June 2, 2024
1
@ColinIanKing Thanks for the fixes!
ps. Tested build and it's clean.
from stress-ng.
ColinIanKing
commented on June 2, 2024
I'm not implicitly using this symbol, the only ppoll symbols in the code are:
objdump -t ./stress-ng | grep ppoll
0000000000710670 l F .text 0000000000000020 bad_ppoll1
0000000000710690 l F .text 0000000000000050 bad_ppoll2
0000000000711140 l F .text 000000000000006a bad_ppoll4
00000000007111b0 l F .text 0000000000000076 bad_ppoll3
0000000000718340 l F .text 00000000000000b7 syscall_ppoll
0000000000000000 F UND 0000000000000000 ppoll
Which distro is this being built on so I can try and reproduce this?
from stress-ng.
vt-alt
commented on June 2, 2024
ALT Linux, there instructions for building in Docker #283 (comment)
There is temporary build log https://git.altlinux.org/tasks/337783/build/200/i586/log
from stress-ng.
vt-alt
commented on June 2, 2024
builder@i586:~/RPM/BUILD/stress-ng-0.17.03$ grep -rl __ppoll_chk
stress-poll.o
stress-ng
builder@i586:~/RPM/BUILD/stress-ng-0.17.04$ objdump -r stress-poll.o | grep __ppoll_chk
0000068c R_386_PLT32 __ppoll_chk
00000773 R_386_PLT32 __ppoll_chk
00000842 R_386_PLT32 __ppoll_chk
Only compilation difference from previous version is
old: cc -pipe -frecord-gcc-switches -Wall -g -O2 -flto=auto -march=i586 -mtune=generic -Wall -Wextra -DVERSION='"0.17.03"' -std=gnu99 -fstack-protector-strong -Werror=format-security -U_FORTIFY_SOURCE -D_FORTIFY_SOURCE=2 -c -o stress-poll.o stress-poll.c
new: cc -pipe -frecord-gcc-switches -Wall -g -O2 -flto=auto -march=i586 -mtune=generic -Wall -Wextra -DVERSION='"0.17.04"' -std=gnu99 -c -o stress-poll.o stress-poll.c
(Btw, We have _FORTIFY_SOURCE=3 by default.)
from stress-ng.
vt-alt
commented on June 2, 2024
So I think this is _FORTIFY_SOURCE related, can this be a glibc bug? @ldv-alt
builder@i586:~/RPM/BUILD/stress-ng-0.17.04$ objdump -dS stress-ng | less
...
__fortify_function __fortified_attr_access (__write_only__, 1, 2) int
ppoll (struct pollfd *__fds, nfds_t __nfds, const struct timespec *__timeout,
const __sigset_t *__ss)
{
return __glibc_fortify (ppoll, __nfds, sizeof (*__fds),
11220a: 58 pop %eax
11220b: 8d 85 70 fe ff ff lea -0x190(%ebp),%eax
112211: ff b5 44 fe ff ff push -0x1bc(%ebp)
112217: 57 push %edi
112218: 89 85 3c fe ff ff mov %eax,-0x1c4(%ebp)
11221e: 50 push %eax
11221f: ff b5 6c fe ff ff push -0x194(%ebp)
112225: ff b5 4c fe ff ff push -0x1b4(%ebp)
11222b: e8 b0 c8 f0 ff call 1eae0 <__ppoll_chk@plt>
...
ts.tv_nsec = 1999999999;
1122ea: c7 85 74 fe ff ff ff 93 35 77 movl $0x773593ff,-0x18c(%ebp)
1122f4: ff b5 44 fe ff ff push -0x1bc(%ebp)
1122fa: ff b5 48 fe ff ff push -0x1b8(%ebp)
112300: ff b5 3c fe ff ff push -0x1c4(%ebp)
112306: ff b5 6c fe ff ff push -0x194(%ebp)
11230c: ff b5 4c fe ff ff push -0x1b4(%ebp)
112312: e8 c9 c7 f0 ff call 1eae0 <__ppoll_chk@plt>
if (UNLIKELY(!g_stress_continue_flag))
...
ts.tv_sec = 0;
1123ac: c7 85 70 fe ff ff 00 00 00 00 movl $0x0,-0x190(%ebp)
1123b6: 83 ec 0c sub $0xc,%esp
ts.tv_nsec = 0;
1123b9: c7 85 74 fe ff ff 00 00 00 00 movl $0x0,-0x18c(%ebp)
1123c3: ff b5 44 fe ff ff push -0x1bc(%ebp)
1123c9: ff b5 48 fe ff ff push -0x1b8(%ebp)
1123cf: ff b5 3c fe ff ff push -0x1c4(%ebp)
1123d5: ff b5 6c fe ff ff push -0x194(%ebp)
1123db: ff b5 4c fe ff ff push -0x1b4(%ebp)
1123e1: e8 fa c6 f0 ff call 1eae0 <__ppoll_chk@plt>
(void)setrlimit(RLIMIT_NOFILE, &old_rlim);
...
from stress-ng.
vt-alt
commented on June 2, 2024
FYI. After switched fortify level from 3 to 2 (with -U_FORTIFY_SOURCE -D_FORTIFY_SOURCE=2 in CFLAGS) the non-LFS functions error disappeared. I think I will stick to this.
from stress-ng.
ColinIanKing
commented on June 2, 2024
Looks like a regression due to commit:
commit c9f29f6
Author: Colin Ian King [email protected]
Date: Sat Dec 23 16:50:05 2023 +0000
Makefile: add cc_supports_flag to simplify CFLAGS checking
from stress-ng.
vt-alt
commented on June 2, 2024
Thanks. With this commit cherry-picked your redefinition of _FORTIFY_SOURCE is returned, as in previous release, and non-LFS functions error disappeared. Except, there is 369 new warnings like this:
[00:00:05] CC core-perf.c
[00:00:05] cc -pipe -frecord-gcc-switches -Wall -g -O2 -flto=auto -march=i586 -mtune=generic -Wall -Wextra -DVERSION='"0.17.04"' -std=gnu99 -Wformat -fstack-protector-strong -D_FORTIFY_SOURCE=2 -U_FORTIFY_SOURCE -Werror=format-security -fipa-pta -c -o core-perf.o core-perf.c
[00:00:05] <command-line>: warning: "_FORTIFY_SOURCE" redefined
[00:00:05] <built-in>: note: this is the location of the previous definition
I think this is because it should be ordered as -U_FORTIFY_SOURCE -D_FORTIFY_SOURCE=2. In commit c9f29f6 ("Makefile: add cc_supports_flag to simplify CFLAGS checking") you also changed their ordering.
from stress-ng.
ColinIanKing
commented on June 2, 2024
Ah, OK, I'll fix that
from stress-ng.
Related Issues (20)
- Execution correctness checks? HOT 1
- stress-ng --io-uring "memory leak" HOT 4
- Request: reorganize the repo so more files are in a subfolder so that the Git page is less onerous to scroll through HOT 2
- sigsegv test failed with Ubuntu T-3.13 HOT 3
- dev test will hang with "task blocked for more than 120 seconds" on Ubuntu T-3.13 HOT 4
- ARM64 Cross Compile Linker Error - implicit pthreads dependency in librt HOT 2
- cache level test /L1/L2 HOT 5
- No stress workers invoked HOT 1
- how to increase cache miss HOT 3
- New version stress-ng, most result of cpu-method test is lower than old version stress-ng HOT 3
- can't be compiled for armv7 HOT 2
- -Werror: <command-line>: error: "VERSION" redefined [-Werror]: using clock_gettime ... no HOT 15
- prio-inv hangs on GCP g1-small with Ubuntu 5.4 / 4.15 / 4.4 lowlatency kernel HOT 4
- stress-ng-0.17.06 cgroup stressor finished subprocess become zombie HOT 2
- stress-ng --io-uring stressor bug HOT 13
- stress-inotify.c:33:46: error: '__s32' undeclared (first use in this function); did you mean '__m32'? HOT 1
- [k8s] How to stress test with 0.5 vcpu on a pod container ? HOT 2
- 0.17.07: build fails HOT 15
- -vm N in command line, but only one worker HOT 4
- Backoff parameter not working HOT 2
Recommend Projects
-
React
A declarative, efficient, and flexible JavaScript library for building user interfaces.
-
Vue.js
🖖 Vue.js is a progressive, incrementally-adoptable JavaScript framework for building UI on the web.
-
Typescript
TypeScript is a superset of JavaScript that compiles to clean JavaScript output.
-
TensorFlow
An Open Source Machine Learning Framework for Everyone
-
Django
The Web framework for perfectionists with deadlines.
-
Laravel
A PHP framework for web artisans
-
D3
Bring data to life with SVG, Canvas and HTML. 📊📈🎉
-
Recommend Topics
-
javascript
JavaScript (JS) is a lightweight interpreted programming language with first-class functions.
-
web
Some thing interesting about web. New door for the world.
-
server
A server is a program made to process requests and deliver data to clients.
-
Machine learning
Machine learning is a way of modeling and interpreting data that allows a piece of software to respond intelligently.
-
Visualization
Some thing interesting about visualization, use data art
-
Game
Some thing interesting about game, make everyone happy.
Recommend Org
-
Facebook
We are working to build community through open source technology. NB: members must have two-factor auth.
-
Microsoft
Open source projects and samples from Microsoft.
-
Google
Google ❤️ Open Source for everyone.
-
Alibaba
Alibaba Open Source for everyone
-
D3
Data-Driven Documents codes.
-
Tencent
China tencent open source team.
from stress-ng.