Messages in this thread Patch in this message | | | Date | Thu, 28 May 2015 10:33:48 +0200 | From | Jiri Olsa <> | Subject | Re: [PATCH v6] perf: __kmod_path__parse: deal with kernel module names in '[]' correctly. |
| |
On Tue, Apr 21, 2015 at 03:33:10AM +0000, Wang Nan wrote: > Before patch ba92732e9808df679ddf75c5ea1c0caae6d7dce2 ('perf kmaps: > Check kmaps to make code more robust'), perf report and perf annotate > will segfault if trace data contains kernel module information like > this:
sry I've lost track on this one.. found it under: [PATCH v4 2/2] perf: report/annotate: fix segfault problem.
my comments were addressed and other than one nit below it looks ok to me
Namhyung, I looked up in history that you had question on this one, could you please recheck Wang's reply?
thanks, jirka
SNIP
> diff --git a/tools/perf/util/dso.c b/tools/perf/util/dso.c > index fc0ddd5..e9d4ae4 100644 > --- a/tools/perf/util/dso.c > +++ b/tools/perf/util/dso.c > @@ -165,13 +165,25 @@ bool is_supported_compression(const char *ext) > return false; > } > > -bool is_kernel_module(const char *pathname) > +bool is_kernel_module(const char *pathname, int cpumode) > { > struct kmod_path m; > > - if (kmod_path__parse(&m, pathname)) > - return NULL; > + /* caller should pass a masked cpumode. Mask again for safety. */ > + switch (cpumode & PERF_RECORD_MISC_CPUMODE_MASK) {
I think it's better to print out if there's unexpected error as warning rather than hide it.. like we do with WARN_ONCE or pr_warning stuff
> + case PERF_RECORD_MISC_USER: > + case PERF_RECORD_MISC_HYPERVISOR: > + case PERF_RECORD_MISC_GUEST_USER: > + return false; > + /* Regard PERF_RECORD_MISC_CPUMODE_UNKNOWN as kernel */ > + default: > + if (kmod_path__parse(&m, pathname)) { > + pr_err("Failed to check whether %s is a kernel module or not. Assume it is.", > + pathname); > > + return true; > + } > + } > return m.kmod;
SNIP
| |