----- Original Message -----
> From: "Cyril Hrubis" <[email protected]>
> To: [email protected]
> Sent: Monday, 10 November, 2014 9:41:08 AM
> Subject: [LTP] LTP usage survey
> 
> Hi!
> I'm working on an article about LTP for lwn.net. It will be a short
> introduction to the project and to the test library. Generally noting
> new for most of you.
> 
> What Jake asked for and I think which will be interesting information
> is where is LTP used and what bugs it's able to find. This is where I
> need your help.
> 
> Please take a few minutes and tell us what do you use LTP for and if
> it's easy enough point to a bugreports/patches/etc.

Hi,

in scope of my team, it's used on daily basis as part of automated
sanity/smoke tests for (RHEL) kernel builds. It consists of subset
of tests, that try to cover as much as possible in limited timeframe
(e.g. <2 hours). kernel QE runs whole sets (runtests), like
various stress and long running tests on release candidates.

Since January 2014, there were 125 BZs reported/updated against
RHEL kernel mentioning "LTP" (case sensitive).

It found all kinds of bugs over last couple years, change in syscall
behaviour, exposed memory leak, crashes or races. Testcases were not
always designed to catch that particular issue, which is where "-i"
parameter comes often handy when trying to debug whether it is kernel
userspace or test problem.

Here are few (upstream) examples I can think of of top of my head:
(greping kernel tree for "LTP" will surely find more)

1. pthread_cond_broadcast revelead issue with futex code:
  commit 69cd9eba38867a493a043bb13eb9b33cad5f1a9a
  Author: Linus Torvalds <[email protected]>
  Date:   Tue Apr 8 15:30:07 2014 -0700
      futex: avoid race between requeue and wake

2. move_pages or migrate_pages found an issue with manual
page migration:
  commit 27329369c9ecf37771b2a65202cbf5578cff3331
  Author: Johannes Weiner <[email protected]>
  Date:   Mon Mar 3 15:38:41 2014 -0800
      mm: page_alloc: exempt GFP_THISNODE allocations from zone fairness

3. mmapstress proved as reliable reproducer for panics, that started
after kernel has been compiled with newer gcc:
  commit b6a9b7f6b1f21735a7456d534dc0e68e61359d2c
  Author: Jan Stancek <[email protected]>
  Date:   Thu Apr 4 11:35:10 2013 -0700
    mm: prevent mmap_cache race in find_vma()

LTP/lcov is also the partial reason, why there was push to get
gcov fixed for new gcc -> nice coverage reports for kernel code
  commit 5f41ea0386a53414d688cfcaa321a78310e5f7c1
  Author: Frantisek Hrbata <[email protected]>
  Date:   Tue Nov 12 15:11:26 2013 -0800
    gcov: add support for gcc 4.7 gcov format

Maybe a lcov report from full LTP run (e.g. x86_64 2+ NUMA system),
on latest stable kernel, showing coverage of various kernel subdirs
would be interesting.

Regards,
Jan

> 
> --
> Cyril Hrubis
> [email protected]
> 
> ------------------------------------------------------------------------------
> _______________________________________________
> Ltp-list mailing list
> [email protected]
> https://lists.sourceforge.net/lists/listinfo/ltp-list
> 

------------------------------------------------------------------------------
Comprehensive Server Monitoring with Site24x7.
Monitor 10 servers for $9/Month.
Get alerted through email, SMS, voice calls or mobile push notifications.
Take corrective actions from your mobile device.
http://pubads.g.doubleclick.net/gampad/clk?id=154624111&iu=/4140/ostg.clktrk
_______________________________________________
Ltp-list mailing list
[email protected]
https://lists.sourceforge.net/lists/listinfo/ltp-list

Reply via email to