From mboxrd@z Thu Jan 1 00:00:00 1970 Content-Type: multipart/mixed; boundary="===============5048372016752564582==" MIME-Version: 1.0 From: Ye Xiaolong To: lkp@lists.01.org Subject: Re: [lkp-robot] [irq/affinity] 13c024422c: fsmark.files_per_sec -4.3% regression Date: Fri, 14 Apr 2017 08:49:07 +0800 Message-ID: <20170414004907.GL16549@yexl-desktop> In-Reply-To: List-Id: --===============5048372016752564582== Content-Type: text/plain; charset="utf-8" MIME-Version: 1.0 Content-Transfer-Encoding: quoted-printable On 04/14, Busch, Keith wrote: >Actually, additional debug is not necessary. I see the accounting error in= the patch you identified and have a fix ready for submission. Thank you fo= r finding the mistake. The incorrect setting required a fairly large number= of CPU's with that NVMe controller, so I wouldn't have noticed it on most = of the servers I have access to. The 0-day regression system is incredibly = helpful! > I am working on to get these debug data while I receive this message, glad = 0day helps :). Thanks, Xiaolong >-----Original Message----- >From: Keith Busch [mailto:keith.busch(a)intel.com] = >Sent: Thursday, April 13, 2017 8:24 AM >To: Ye, Xiaolong >Cc: lkp(a)01.org >Subject: Re: [lkp-robot] [irq/affinity] 13c024422c: fsmark.files_per_sec -= 4.3% regression > >On Thu, Apr 13, 2017 at 09:06:04AM +0800, Ye Xiaolong wrote: >> On 04/12, Busch, Keith wrote: >> >Hi Xiaolong, >> > >> >Thanks for letting me know of the regression. This change wasn't suppos= ed to affect anything outside the unbalanced numa case, so if it's indeed t= he culprit, I'll need to fix that. >> > >> >From the machine that demonstrates the regression, could you send me 'l= scpu'? And the output from the following commands, both with and without th= e patch applied? >> = >> Okay, when should I run the `lscpu` and below cmds, before, after or dur= ing the test? > >The lscpu shouldn't change at any time, so you can grab that either >before or after, with any set of patches applied. --===============5048372016752564582==--