[Bug translator/17055] _stp_perf_read needs a sleepable context

["dsmith at redhat dot com" <sourceware-bugzilla at sourceware dot org>]

https://sourceware.org/bugzilla/show_bug.cgi?id=17055

--- Comment #2 from David Smith <dsmith at redhat dot com> ---
Created attachment 9237
  --> https://sourceware.org/bugzilla/attachment.cgi?id=9237&action=edit
rough patch

Here's a rough patch that tries to fix this problem. The tricky part here is
that we now read the perf counters before we've got a context structure. For
storage, I'm using a per-cpu variable.

With this patch, all the perf tests pass, except for the last subtest in
perf.exp. In that subtest, it runs a script twice, the 2nd time swapping the
order of declarations of the perf counters. When this is run with the new code,
the subtest fails because the values returned are different between the 2 runs.
I don't know what is going on there.

There are several "FIXME" comments scattered throughout the patch that need
reviewing.

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