[Bug runtime/29756] New: stap -L regexp match broken on some arches

public inbox for systemtap@sourceware.org
 help / color / mirror / Atom feed

* [Bug runtime/29756] New: stap -L regexp match broken on some arches
@ 2022-11-07 10:35 mcermak at redhat dot com
  2023-10-12 20:16 ` [Bug runtime/29756] " wcohen at redhat dot com
                   ` (2 more replies)
  0 siblings, 3 replies; 4+ messages in thread
From: mcermak at redhat dot com @ 2022-11-07 10:35 UTC (permalink / raw)
  To: systemtap

https://sourceware.org/bugzilla/show_bug.cgi?id=29756

            Bug ID: 29756
           Summary: stap -L regexp match broken on some arches
           Product: systemtap
           Version: unspecified
            Status: NEW
          Severity: normal
          Priority: P2
         Component: runtime
          Assignee: systemtap at sourceware dot org
          Reporter: mcermak at redhat dot com
  Target Milestone: ---

On some arches (interestingly, this only seems to be a problem on ppc64le and
s390x, but not on x86_64 or aarch64) the stap -L regexp match doesn't work for
a function name in case the source file isn't specified.  If the source file is
specified (a base name without any kind of path is good enough) then the match
works fine. Details:

9 ppc64le # stap -L 'process("/lib64/libc.so.6").function("malloc_init_state")'
9 ppc64le # stap -L
'process("/lib64/libc.so.6").function("malloc_init_state*")'
process("/usr/lib64/libc.so.6").function("malloc_init_state@/usr/src/debug/glibc-2.34-48.el9.ppc64le/malloc/malloc.c:1932")
$av:mstate
9 ppc64le # stap -L
'process("/lib64/libc.so.6").function("malloc_init_state@/usr/src/debug/glibc-2.34-48.el9.ppc64le/malloc/malloc.c")'
process("/usr/lib64/libc.so.6").function("malloc_init_state@/usr/src/debug/glibc-2.34-48.el9.ppc64le/malloc/malloc.c:1932")
$av:mstate
9 ppc64le # stap -L
'process("/lib64/libc.so.6").function("malloc_init_state@malloc.c")'
process("/usr/lib64/libc.so.6").function("malloc_init_state@/usr/src/debug/glibc-2.34-48.el9.ppc64le/malloc/malloc.c:1932")
$av:mstate
9 ppc64le # stap -L 'process("/lib64/libc.so.6").function("malloc_init_state")'
9 ppc64le #

This is with  kernel 5.14.0-186.kpq1.el9.ppc64le and stap upstream commit
e51e989e33b0f0a0cf26d00755340cef3c2ea81f .

-- 
You are receiving this mail because:
You are the assignee for the bug.

^ permalink raw reply	[flat|nested] 4+ messages in thread

* [Bug runtime/29756] stap -L regexp match broken on some arches
  2022-11-07 10:35 [Bug runtime/29756] New: stap -L regexp match broken on some arches mcermak at redhat dot com
@ 2023-10-12 20:16 ` wcohen at redhat dot com
  2023-10-13 19:24 ` wcohen at redhat dot com
  2023-10-16 19:58 ` [Bug translator/29756] " wcohen at redhat dot com
  2 siblings, 0 replies; 4+ messages in thread
From: wcohen at redhat dot com @ 2023-10-12 20:16 UTC (permalink / raw)
  To: systemtap

https://sourceware.org/bugzilla/show_bug.cgi?id=29756

William Cohen <wcohen at redhat dot com> changed:

           What    |Removed                     |Added
----------------------------------------------------------------------------
                 CC|                            |wcohen at redhat dot com

--- Comment #1 from William Cohen <wcohen at redhat dot com> ---
I am able to replicates this on s390x RHEL9 and a build of the current git
checkout of systemtap commit 3ecf49eee2ab7e7:

[root@s390x-kvm-032 systemtap]# rpm -q kernel
kernel-5.14.0-362.8.1.el9_3.s390x
[root@s390x-kvm-032 systemtap]# ../install/bin/stap -L
'process("/lib64/libc.so.6").function("malloc_init_state")'
[root@s390x-kvm-032 systemtap]# ../install/bin/stap -L
'process("/lib64/libc.so.6").function("malloc_init_state*")'
process("/usr/lib64/libc.so.6").function("malloc_init_state@/usr/src/debug/glibc-2.34-83.el9_3.7.s390x/malloc/malloc.c:1932")
$av:mstate $tcache:tcache_perthread_struct*

On a similar s390x RHEL8 system with the same checkout don't see this issue:

[root@s390x-kvm-003 systemtap]#  rpm -q kernel
kernel-4.18.0-372.9.1.el8.s390x
[root@s390x-kvm-003 systemtap]# ../install/bin/stap -L
'process("/lib64/libc.so.6").function("malloc_init_state")'
process("/usr/lib64/libc-2.28.so").function("malloc_init_state@/usr/src/debug/glibc-2.28-189.1.el8.s390x/malloc/malloc.c:1803")
$av:mstate $i:int
[root@s390x-kvm-003 systemtap]#  ../install/bin/stap -L
'process("/lib64/libc.so.6").function("malloc_init_state*")'
process("/usr/lib64/libc-2.28.so").function("malloc_init_state@/usr/src/debug/glibc-2.28-189.1.el8.s390x/malloc/malloc.c:1803")
$av:mstate $i:int

-- 
You are receiving this mail because:
You are the assignee for the bug.

^ permalink raw reply	[flat|nested] 4+ messages in thread

* [Bug runtime/29756] stap -L regexp match broken on some arches
  2022-11-07 10:35 [Bug runtime/29756] New: stap -L regexp match broken on some arches mcermak at redhat dot com
  2023-10-12 20:16 ` [Bug runtime/29756] " wcohen at redhat dot com
@ 2023-10-13 19:24 ` wcohen at redhat dot com
  2023-10-16 19:58 ` [Bug translator/29756] " wcohen at redhat dot com
  2 siblings, 0 replies; 4+ messages in thread
From: wcohen at redhat dot com @ 2023-10-13 19:24 UTC (permalink / raw)
  To: systemtap

https://sourceware.org/bugzilla/show_bug.cgi?id=29756

--- Comment #2 from William Cohen <wcohen at redhat dot com> ---
Did some more poking around to see what the differences are between the rhel8
and rhel9 libc.so.6 code.  It appears that the problem is linked to functions
being inlined on rhel9.

On rhel9 the particular function being searched for is inlined while on rhel8
it is an actual function.  RHEL9:

# ../install/bin/stap -v  -L
'process("/lib64/libc.so.6").function("malloc_init_state*").*'
Pass 1: parsed user script and 489 library scripts using
127816virt/106228res/10964shr/94752data kb, in 240usr/40sys/286real ms.
process("/usr/lib64/libc.so.6").function("malloc_init_state@/usr/src/debug/glibc-2.34-83.el9_3.7.s390x/malloc/malloc.c:1932").inline
/* pc=.dynamic+0xab576 */ $av:mstate $tcache:tcache_perthread_struct*
process("/usr/lib64/libc.so.6").function("malloc_init_state@/usr/src/debug/glibc-2.34-83.el9_3.7.s390x/malloc/malloc.c:1932").inline
/* pc=.dynamic+0xab9e0 */ $av:mstate $tcache:tcache_perthread_struct*
Pass 2: analyzed script: 2 probes, 0 functions, 0 embeds, 0 globals using
152984virt/132752res/12260shr/119920data kb, in 130usr/10sys/142real ms.

0xab576 is inlined in arena_get2.part.0
0xab9e0 is inlined in ptmalloc_init.part.0

on rhel8 see .call and .return:
# ../install/bin/stap -v  -L
'process("/lib64/libc.so.6").function("malloc_init_state*").*'
Pass 1: parsed user script and 480 library scripts using
122520virt/102760res/12516shr/89836data kb, in 230usr/20sys/248real ms.
process("/usr/lib64/libc-2.28.so").function("malloc_init_state@/usr/src/debug/glibc-2.28-189.1.el8.s390x/malloc/malloc.c:1803").call
/* pc=.dynamic+0x92870 */ $av:mstate $i:int
process("/usr/lib64/libc-2.28.so").function("malloc_init_state@/usr/src/debug/glibc-2.28-189.1.el8.s390x/malloc/malloc.c:1803").return
/* pc=.dynamic+0x92870 */ $av:mstate $i:int
Pass 2: analyzed script: 2 probes, 0 functions, 0 embeds, 0 globals using
143736virt/125236res/13712shr/111052data kb, in 110usr/10sys/124real ms.

When trying "stap -v -L ..." on rhel9 libc.so.6 functions that were regular
functions no trailing '*' was required.  However, for inlined functions needed
the '*' to get them listed out.  For example ptmalloc_init.

[root@s390x-kvm-032 systemtap]# ../install/bin/stap -v  -L
'process("/lib64/libc.so.6").function("ptmalloc_init").inline'
Pass 1: parsed user script and 489 library scripts using
127812virt/106224res/10964shr/94748data kb, in 230usr/30sys/309real ms.
Pass 2: analyzed script: 0 probes, 0 functions, 0 embeds, 0 globals using
152848virt/132860res/12452shr/119784data kb, in 110usr/10sys/124real ms.
[root@s390x-kvm-032 systemtap]# ../install/bin/stap -v  -L
'process("/lib64/libc.so.6").function("ptmalloc_init*").inline'
Pass 1: parsed user script and 489 library scripts using
127816virt/106228res/10964shr/94752data kb, in 230usr/30sys/274real ms.
process("/usr/lib64/libc.so.6").function("ptmalloc_init@/usr/src/debug/glibc-2.34-83.el9_3.7.s390x/malloc/arena.c:281").inline
/* pc=.dynamic+0xb02b4 */ $tcache:tcache_perthread_struct*
process("/usr/lib64/libc.so.6").function("ptmalloc_init@/usr/src/debug/glibc-2.34-83.el9_3.7.s390x/malloc/arena.c:281").inline
/* pc=.dynamic+0xb0210 */ $tcache:tcache_perthread_struct*
process("/usr/lib64/libc.so.6").function("ptmalloc_init@/usr/src/debug/glibc-2.34-83.el9_3.7.s390x/malloc/arena.c:281").inline
/* pc=.dynamic+0xaffa0 */ $tcache:tcache_perthread_struct*
process("/usr/lib64/libc.so.6").function("ptmalloc_init@/usr/src/debug/glibc-2.34-83.el9_3.7.s390x/malloc/arena.c:281").inline
/* pc=.dynamic+0xafdf8 */ $tcache:tcache_perthread_struct*
process("/usr/lib64/libc.so.6").function("ptmalloc_init@/usr/src/debug/glibc-2.34-83.el9_3.7.s390x/malloc/arena.c:281").inline
/* pc=.dynamic+0xaf950 */ $tcache:tcache_perthread_struct*
process("/usr/lib64/libc.so.6").function("ptmalloc_init@/usr/src/debug/glibc-2.34-83.el9_3.7.s390x/malloc/arena.c:281").inline
/* pc=.dynamic+0xaf71e */ $tcache:tcache_perthread_struct*
process("/usr/lib64/libc.so.6").function("ptmalloc_init@/usr/src/debug/glibc-2.34-83.el9_3.7.s390x/malloc/arena.c:281").inline
/* pc=.dynamic+0xaf6aa */ $tcache:tcache_perthread_struct*
process("/usr/lib64/libc.so.6").function("ptmalloc_init@/usr/src/debug/glibc-2.34-83.el9_3.7.s390x/malloc/arena.c:281").inline
/* pc=.dynamic+0xaf656 */ $tcache:tcache_perthread_struct*
process("/usr/lib64/libc.so.6").function("ptmalloc_init@/usr/src/debug/glibc-2.34-83.el9_3.7.s390x/malloc/arena.c:281").inline
/* pc=.dynamic+0xaf31c */ $tcache:tcache_perthread_struct*
process("/usr/lib64/libc.so.6").function("ptmalloc_init@/usr/src/debug/glibc-2.34-83.el9_3.7.s390x/malloc/arena.c:281").inline
/* pc=.dynamic+0xaea7c */ $tcache:tcache_perthread_struct*
process("/usr/lib64/libc.so.6").function("ptmalloc_init@/usr/src/debug/glibc-2.34-83.el9_3.7.s390x/malloc/arena.c:281").inline
/* pc=.dynamic+0xb04a2 */ $tcache:tcache_perthread_struct*
process("/usr/lib64/libc.so.6").function("ptmalloc_init@/usr/src/debug/glibc-2.34-83.el9_3.7.s390x/malloc/arena.c:281").inline
/* pc=.dynamic+0xb0504 */ $tcache:tcache_perthread_struct*
Pass 2: analyzed script: 12 probes, 0 functions, 0 embeds, 0 globals using
152984virt/132764res/12260shr/119920data kb, in 120usr/10sys/129real ms.

-- 
You are receiving this mail because:
You are the assignee for the bug.

^ permalink raw reply	[flat|nested] 4+ messages in thread

* [Bug translator/29756] stap -L regexp match broken on some arches
  2022-11-07 10:35 [Bug runtime/29756] New: stap -L regexp match broken on some arches mcermak at redhat dot com
  2023-10-12 20:16 ` [Bug runtime/29756] " wcohen at redhat dot com
  2023-10-13 19:24 ` wcohen at redhat dot com
@ 2023-10-16 19:58 ` wcohen at redhat dot com
  2 siblings, 0 replies; 4+ messages in thread
From: wcohen at redhat dot com @ 2023-10-16 19:58 UTC (permalink / raw)
  To: systemtap

https://sourceware.org/bugzilla/show_bug.cgi?id=29756

William Cohen <wcohen at redhat dot com> changed:

           What    |Removed                     |Added
----------------------------------------------------------------------------
          Component|runtime                     |translator

--- Comment #3 from William Cohen <wcohen at redhat dot com> ---
This looks like an issue with systemtap understanding the debuginfo generated
as part of the rpmbuild process.  When locally building the
glibc-2.34-83.el9_3.7.src.rpm on the machine the following work fine on the
unstripped glibc.so.6:

[root@s390x-kvm-033 build-s390x-redhat-linux]# 
~/systemtap_write/install/bin/stap -v -L
'process("./libc.so.6").function("malloc_init_state")'
Pass 1: parsed user script and 489 library scripts using
127764virt/106232res/10972shr/94700data kb, in 230usr/30sys/256real ms.
process("/root/rpmbuild/BUILD/glibc-2.34/build-s390x-redhat-linux/libc.so").function("malloc_init_state@/root/rpmbuild/BUILD/glibc-2.34/malloc/malloc.c:1932")
/* pc=.dynamic+0xab576 */ $av:mstate
process("/root/rpmbuild/BUILD/glibc-2.34/build-s390x-redhat-linux/libc.so").function("malloc_init_state@/root/rpmbuild/BUILD/glibc-2.34/malloc/malloc.c:1932")
/* pc=.dynamic+0xab9e0 */ $av:mstate
Pass 2: analyzed script: 2 probes, 0 functions, 0 embeds, 0 globals using
137564virt/117272res/12220shr/104500data kb, in 50usr/0sys/57real ms.

However, install the rpms created from that same build a wildcard is needed to
find malloc_init_state:

#cd rpmbuild/RPMS/s390x/
#dnf install -y ./glibc-gconv-extra-2.34-83.el9.7.s390x.rpm 
./glibc-langpack-en-2.34-83.el9.7.s390x.rpm
./glibc-common-2.34-83.el9.7.s390x.rpm ./glibc-2.34-83.el9.7.s390x.rpm 
./glibc-headers-2.34-83.el9.7.s390x.rpm ./glibc-devel-2.34-83.el9.7.s390x.rpm
./glibc-debugsource-2.34-83.el9.7.s390x.rpm
./glibc-debuginfo-2.34-83.el9.7.s390x.rpm
./glibc-static-2.34-83.el9.7.s390x.rpm 

#  ~/systemtap_write/install/bin/stap -v -L
'process("/lib64/libc.so.6").function("malloc_init_state")'
Pass 1: parsed user script and 489 library scripts using
127764virt/106232res/10972shr/94700data kb, in 230usr/30sys/259real ms.
Pass 2: analyzed script: 0 probes, 0 functions, 0 embeds, 0 globals using
151728virt/131664res/12412shr/118664data kb, in 100usr/0sys/113real ms.
#  ~/systemtap_write/install/bin/stap -v -L
'process("/lib64/libc.so.6").function("malloc_init_state*")'
Pass 1: parsed user script and 489 library scripts using
127768virt/106236res/10972shr/94704data kb, in 240usr/30sys/254real ms.
process("/usr/lib64/libc.so.6").function("malloc_init_state@/usr/src/debug/glibc-2.34-83.el9.7.s390x/malloc/malloc.c:1932")
/* pc=.dynamic+0xab576 */ $av:mstate $tcache:tcache_perthread_struct*
process("/usr/lib64/libc.so.6").function("malloc_init_state@/usr/src/debug/glibc-2.34-83.el9.7.s390x/malloc/malloc.c:1932")
/* pc=.dynamic+0xab9e0 */ $av:mstate $tcache:tcache_perthread_struct*
Pass 2: analyzed script: 2 probes, 0 functions, 0 embeds, 0 globals using
151820virt/131536res/12220shr/118756data kb, in 110usr/0sys/120real ms.

-- 
You are receiving this mail because:
You are the assignee for the bug.

^ permalink raw reply	[flat|nested] 4+ messages in thread

end of thread, other threads:[~2023-10-16 19:58 UTC | newest]

Thread overview: 4+ messages (download: mbox.gz / follow: Atom feed)
-- links below jump to the message on this page --
2022-11-07 10:35 [Bug runtime/29756] New: stap -L regexp match broken on some arches mcermak at redhat dot com
2023-10-12 20:16 ` [Bug runtime/29756] " wcohen at redhat dot com
2023-10-13 19:24 ` wcohen at redhat dot com
2023-10-16 19:58 ` [Bug translator/29756] " wcohen at redhat dot com

This is a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox;
as well as URLs for read-only IMAP folder(s) and NNTP newsgroup(s).