linux-kernel - Re: perf: regression with PERF_EVENT_IOC

lists.openwall.net		lists / announce owl-users owl-dev john-users john-dev passwdqc-users yescrypt popa3d-users / oss-security kernel-hardening musl sabotage tlsify passwords / crypt-dev xvendor / Bugtraq Full-Disclosure linux-kernel linux-netdev linux-ext4 linux-hardening linux-cve-announce PHC
Open Source and information security mailing list archives

Hash Suite: Windows password security audit tool. GUI, reports in PDF.

[<prev] [next>] [<thread-prev] [thread-next>] [day] [month] [year] [list]

Date:	Tue, 24 May 2011 17:40:20 +0200
From:	Ingo Molnar <mingo@...e.hu>
To:	Peter Zijlstra <a.p.zijlstra@...llo.nl>
Cc:	Vince Weaver <vweaver1@...s.utk.edu>, linux-kernel@...r.kernel.org,
	fbuihuu@...il.com, paulus@...ba.org, acme@...hat.com
Subject: Re: perf: regression with PERF_EVENT_IOC_REFRESH

* Peter Zijlstra <a.p.zijlstra@...llo.nl> wrote:

> On Tue, 2011-05-24 at 11:04 -0400, Vince Weaver wrote:
>
> > 2.6.37, 2.6.38, or 2.6.39 then it would be silly to do it just for
> > 2.6.40.

No, this commit was added in v2.6.38 so v2.6.37 should be fine.

> Oh, I assumed it was recent and .39/.40 would suffice.

Btw., how did it happen that the PAPI tests did not get run against upstream 
over the course of about half a year, two full stable kernels released:

 Date:   Mon Mar 14 18:20:32 2011 -0700    Linux 2.6.38
 Date:   Wed May 18 21:06:34 2011 -0700    Linux 2.6.39

?

I'd suggest periodically running the PAPI tests on the perf development tree:

  http://people.redhat.com/mingo/tip.git/README

doing that would have caught this problem 6 months ago.

The upstream policy is that regressions are generally recognized before the 
next kernel gets released: i.e. in the stabilization period after -rc1, the 
roughly two months until the final kernel gets released. That is the window 
when we can still fix regressions relatively cheaply.

Yes, there are exceptions, but if a piece of user-space code did not get tested 
with upstream over months and months then that moves into the 'fix it if we 
can' category - not a regression per se.

So the upstream message is: we can only care about you if you care testing 
upstream.

So if it's easy to fix we can certainly fix this bug and mark it for a -stable 
backport, but this is not a regression that got reported to us in any timely 
manner.

Btw., to get such assumptions tested more frequently than twice a year i'd 
suggest moving these usecases into 'perf test' or so - that it gets run every 
day:

 $ perf test
 1: vmlinux symtab matches kallsyms: FAILED!

 2: detect open syscall event: Ok
 3: detect open syscall event on all cpus: Ok
 4: read samples using the mmap interface: Ok

Thanks,

	Ingo
--
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to majordomo@...r.kernel.org
More majordomo info at  http://vger.kernel.org/majordomo-info.html
Please read the FAQ at  http://www.tux.org/lkml/