Bug 217038
Summary: | Latest trace-cmd-report seems to ignores filter (-F) parameter | ||
---|---|---|---|
Product: | Tools | Reporter: | Gabriel Krisman Bertazi (krisman) |
Component: | Trace-cmd/Kernelshark | Assignee: | Default virtual assignee for Trace-cmd and kernelshark (tools_tracecmd_kernelshark) |
Status: | RESOLVED CODE_FIX | ||
Severity: | normal | CC: | rostedt |
Priority: | P1 | ||
Hardware: | All | ||
OS: | Linux | ||
Kernel Version: | 5.14.21-150400.24.38-default | Subsystem: | |
Regression: | No | Bisected commit-id: | |
Attachments: | patch on the ML. |
Description
Gabriel Krisman Bertazi
2023-02-14 22:09:46 UTC
Strange? I just tried it out (with the latest of all the libraries too) # trace-cmd report -F mm_page_alloc -v -F kernel_stack cpus=4 trace-cmd-2383 [000] 7192.837686: mm_page_alloc: page=0xffffffff936e3208 pfn=0x25658 order=0 migratetype=1 gfp_flags=GFP_HIGHUSER_MOVABLE trace-cmd-2386 [002] 7192.837698: mm_page_alloc: page=0xffffffff936ec67f pfn=0x2eacf order=0 migratetype=1 gfp_flags=GFP_HIGHUSER_MOVABLE trace-cmd-2386 [002] 7192.837735: mm_page_alloc: page=0xffffffff936ecb10 pfn=0x2ef60 order=0 migratetype=1 gfp_flags=GFP_HIGHUSER_MOVABLE trace-cmd-2386 [002] 7192.837754: mm_page_alloc: page=0xffffffff936ecc05 pfn=0x2f055 order=0 migratetype=1 gfp_flags=GFP_HIGHUSER_MOVABLE trace-cmd-2386 [002] 7192.837759: mm_page_alloc: page=0xffffffff936ecb2e pfn=0x2ef7e order=0 migratetype=1 gfp_flags=GFP_HIGHUSER_MOVABLE trace-cmd-2386 [002] 7192.837763: mm_page_alloc: page=0xffffffff936ecb2f pfn=0x2ef7f order=0 migratetype=1 gfp_flags=GFP_HIGHUSER_MOVABLE trace-cmd-2386 [002] 7192.837768: mm_page_alloc: page=0xffffffff936ec723 pfn=0x2eb73 order=0 migratetype=1 gfp_flags=GFP_HIGHUSER_MOVABLE trace-cmd-2386 [002] 7192.837782: mm_page_alloc: page=0xffffffff936c3110 pfn=0x5560 order=2 migratetype=0 gfp_flags=GFP_KERNEL|__GFP_NOWARN|__GFP_NORETRY|__GFP_COMP|__GFP_ZERO trace-cmd-2386 [002] 7192.837794: mm_page_alloc: page=0xffffffff936e1f2a pfn=0x2437a order=0 migratetype=1 gfp_flags=GFP_HIGHUSER_MOVABLE trace-cmd-2386 [002] 7192.837804: mm_page_alloc: page=0xffffffff936ecc6a pfn=0x2f0ba order=0 migratetype=1 gfp_flags=GFP_HIGHUSER_MOVABLE trace-cmd-2387 [001] 7192.837816: mm_page_alloc: page=0xffffffff936f894c pfn=0x3ad9c order=2 migratetype=0 gfp_flags=GFP_KERNEL|__GFP_NOWARN|__GFP_NORETRY|__GFP_COMP|__GFP_ZERO trace-cmd-2387 [001] 7192.837847: mm_page_alloc: page=0xffffffff936ecfc3 pfn=0x2f413 order=0 migratetype=1 gfp_flags=GFP_HIGHUSER_MOVABLE trace-cmd-2386 [002] 7192.837848: mm_page_alloc: page=0xffffffff936ec5ca pfn=0x2ea1a order=0 migratetype=1 gfp_flags=GFP_HIGHUSER_MOVABLE trace-cmd-2387 [001] 7192.837859: mm_page_alloc: page=0xffffffff936e00f6 pfn=0x22546 order=0 migratetype=1 gfp_flags=GFP_HIGHUSER_MOVABLE trace-cmd-2386 [002] 7192.837863: mm_page_alloc: page=0xffffffff936e77a8 pfn=0x29bf8 order=0 migratetype=1 gfp_flags=GFP_HIGHUSER_MOVABLE trace-cmd-2387 [001] 7192.837904: mm_page_alloc: page=0xffffffff936e3629 pfn=0x25a79 order=0 migratetype=1 gfp_flags=GFP_HIGHUSER_MOVABLE <idle>-0 [001] 7193.634294: mm_page_alloc: page=0xffffffff936dfa18 pfn=0x21e68 order=3 migratetype=0 gfp_flags=GFP_ATOMIC|__GFP_NOWARN|__GFP_NORETRY|__GFP_COMP kworker/3:2-2327 [003] 7194.749516: mm_page_alloc: page=0xffffffff936c3a06 pfn=0x5e56 order=0 migratetype=0 gfp_flags=GFP_NOFS|__GFP_NOWARN|__GFP_NORETRY <idle>-0 [001] 7195.914313: mm_page_alloc: page=0xffffffff936c0ba0 pfn=0x2ff0 order=3 migratetype=0 gfp_flags=GFP_ATOMIC|__GFP_NOWARN|__GFP_NORETRY|__GFP_COMP <idle>-0 [001] 7195.914326: mm_page_alloc: page=0xffffffff936f8b78 pfn=0x3afc8 order=3 migratetype=0 gfp_flags=GFP_ATOMIC|__GFP_NOWARN|__GFP_NORETRY|__GFP_COMP [root@fedora-vmware trace-cmd.git]# trace-cmd -v trace-cmd version 3.1.6 (0a68daed9f58e4429c0f1e7818f7cc0634873112) # trace-cmd report -F mm_page_alloc | head cpus=4 trace-cmd-2383 [000] 7192.837686: mm_page_alloc: page=0xffffffff936e3208 pfn=0x25658 order=0 migratetype=1 gfp_flags=GFP_HIGHUSER_MOVABLE trace-cmd-2386 [002] 7192.837698: mm_page_alloc: page=0xffffffff936ec67f pfn=0x2eacf order=0 migratetype=1 gfp_flags=GFP_HIGHUSER_MOVABLE trace-cmd-2386 [002] 7192.837726: kernel_stack: <stack trace > => trace_event_raw_event_mm_page_alloc (ffffffff922ada56) => __alloc_pages (ffffffff922e61ad) => alloc_pages_vma (ffffffff92304730) => wp_page_copy (ffffffff922bc5e5) => __handle_mm_fault (ffffffff922c3c03) => handle_mm_fault (ffffffff922c442f) I can't reproduce the issue. Thanks for the prompt response, Steven!
> I can't reproduce the issue.
Indeed, quite strange. This was reported to me on Tumbleweed and I could reproduce it consistently with the builds I mentioned above.
I reported to check with you first if it rings a bell. But let me fire up GDB in the next days and see if I collect more information or figure out what is wrong in the affected systems.
(In reply to Steven Rostedt from comment #1) > Strange? I just tried it out (with the latest of all the libraries too) > > # trace-cmd report -F mm_page_alloc -v -F kernel_stack Hey Steven, Sorry, I could only get back to this today. I just found the problem. It is actually a bit silly: This works fine: tracecmd/trace-cmd report -v -F kernel_stack -i trace.dat This doesn't: tracecmd/trace-cmd report -i trace.dat -v -F kernel_stack Note how the filter needs to come before the input file. You couldn't reproduce because you didn't pass a data file with -i. The argument ordering wasn't required on previous versions. I bisected it to: 955d05fc7a ("trace-cmd report: Make filter arguments match their files") Which make a lot of sense. The commit message suggests this is unintended behavior. I've sent you a patch on the ML. Created attachment 303977 [details]
patch on the ML.
for completeness, the patch i sent.
|