From mboxrd@z Thu Jan 1 00:00:00 1970 Received: from eggs.gnu.org ([2001:4830:134:3::10]:33051) by lists.gnu.org with esmtp (Exim 4.71) (envelope-from ) id 1bGIFZ-0008FI-2q for qemu-devel@nongnu.org; Thu, 23 Jun 2016 23:57:53 -0400 Received: from Debian-exim by eggs.gnu.org with spam-scanned (Exim 4.71) (envelope-from ) id 1bGIFS-0003ig-OT for qemu-devel@nongnu.org; Thu, 23 Jun 2016 23:57:48 -0400 Received: from mail-qk0-x232.google.com ([2607:f8b0:400d:c09::232]:33692) by eggs.gnu.org with esmtp (Exim 4.71) (envelope-from ) id 1bGIFS-0003ic-J9 for qemu-devel@nongnu.org; Thu, 23 Jun 2016 23:57:42 -0400 Received: by mail-qk0-x232.google.com with SMTP id a186so132102017qkf.0 for ; Thu, 23 Jun 2016 20:57:42 -0700 (PDT) Sender: Richard Henderson References: <1456252389-4416-1-git-send-email-rth@twiddle.net> <1456252389-4416-6-git-send-email-rth@twiddle.net> <57607C9C.8070404@ilande.co.uk> <57631FC9.1040303@ilande.co.uk> From: Richard Henderson Message-ID: <65a57d82-174f-cc1c-d613-984bffce13d3@twiddle.net> Date: Thu, 23 Jun 2016 20:57:39 -0700 MIME-Version: 1.0 In-Reply-To: <57631FC9.1040303@ilande.co.uk> Content-Type: text/plain; charset=windows-1252; format=flowed Content-Transfer-Encoding: 7bit Subject: Re: [Qemu-devel] [PULL 5/8] target-sparc: Use global registers for the register window List-Id: List-Unsubscribe: , List-Archive: List-Post: List-Help: List-Subscribe: , To: Mark Cave-Ayland Cc: qemu-devel@nongnu.org On 06/16/2016 02:53 PM, Mark Cave-Ayland wrote: > On 16/06/16 21:26, Richard Henderson wrote: > >> On 06/14/2016 02:52 PM, Mark Cave-Ayland wrote: >>> Following up the bug report at >>> https://bugs.launchpad.net/qemu/+bug/1588328, I bisected the regression >>> down to this particular commit. I can't see anything obvious here, so >>> perhaps this is exposing another bug somewhere else? >>> >> >> Probably. I'm downloading the solaris image now. >> >> >> r~ > > Thanks for taking a look - otherwise I won't be able to get to this > until next week. My thinking was that since the code makes access to > regwptr direct instead of copied into a temporary, something is > accidentally clobbering a destination register... I've been unable to find this. Whatever happens, it happens after 10GB of logs, which is simply too much to sift through. I've tried to narrow it down, but the lack of a hardware tlb refill means that we get hundreds of thousands of Data Access Faults that are simply TLB misses and not the actual Segmentation Fault in question. It doesn't seem to affect other OSes, so I can't imagine what quirk is being exercised in this case. As loath as I am to suggest it, we may have to revert the sparc indirect register patch for the release. I do now ping the rest of my sparc improvements patchset. It's completely independent of the use of indirect registers. r~