|
From: Julian S. <js...@ac...> - 2015-04-09 13:49:11
|
Bug 339778 (https://bugs.kde.org/show_bug.cgi?id=339778) contains patches for a port to TileGx, a 64 bit CPU. There has been quite some reviewing and re-working of the patches. From my point of view they are now ready to land. Are there any more comments or concerns regarding this bug? If so please say so now. If not I propose to land it in the next day or so. J |
|
From: Florian K. <fl...@ei...> - 2015-04-09 16:46:41
|
On 09.04.2015 15:48, Julian Seward wrote: > > Bug 339778 (https://bugs.kde.org/show_bug.cgi?id=339778) contains > patches for a port to TileGx, a 64 bit CPU. There has been quite > some reviewing and re-working of the patches. From my point of > view they are now ready to land. > > Are there any more comments or concerns regarding this bug? If so > please say so now. If not I propose to land it in the next day or so. This looks all pretty good to me. I added a few comments to the BZ. All minor stuff except the absence of insn tests. One thing I'd like to bring up is a nightly build with results posted to the mailing list. We already have enough ports without those and generally no idea how well those ports work and/or how they brake when new stuff gets added. We want to be able to observe that. Florian |
|
From: Roland M. <rol...@nr...> - 2015-04-09 16:55:27
|
On Thu, Apr 9, 2015 at 6:46 PM, Florian Krohm <fl...@ei...> wrote: > On 09.04.2015 15:48, Julian Seward wrote: >> >> Bug 339778 (https://bugs.kde.org/show_bug.cgi?id=339778) contains >> patches for a port to TileGx, a 64 bit CPU. There has been quite >> some reviewing and re-working of the patches. From my point of >> view they are now ready to land. >> >> Are there any more comments or concerns regarding this bug? If so >> please say so now. If not I propose to land it in the next day or so. > > This looks all pretty good to me. > I added a few comments to the BZ. All minor stuff except the absence of > insn tests. > > One thing I'd like to bring up is a nightly build with results posted to > the mailing list. We already have enough ports without those and > generally no idea how well those ports work and/or how they brake when > new stuff gets added. We want to be able to observe that. Is there any (free) emulator software for TileGX which could fill that testing gap somehow ? ---- Bye, Roland -- __ . . __ (o.\ \/ /.o) rol...@nr... \__\/\/__/ MPEG specialist, C&&JAVA&&Sun&&Unix programmer /O /==\ O\ TEL +49 641 3992797 (;O/ \/ \O;) |
|
From: Zhigang L. <zl...@ez...> - 2015-04-10 02:32:03
|
________________________________________ From: gi...@gm... <gi...@gm...> on behalf of Roland Mainz <rol...@nr...> Sent: Thursday, April 9, 2015 12:55 PM To: Valgrind Developers Cc: Julian Seward; Zhigang Liu; Florian Krohm Subject: Re: [Valgrind-developers] Linux/TileGX port: last call for comments On Thu, Apr 9, 2015 at 6:46 PM, Florian Krohm <fl...@ei...> wrote: > On 09.04.2015 15:48, Julian Seward wrote: >> >> Bug 339778 (https://bugs.kde.org/show_bug.cgi?id=339778) contains >> patches for a port to TileGx, a 64 bit CPU. There has been quite >> some reviewing and re-working of the patches. From my point of >> view they are now ready to land. >> >> Are there any more comments or concerns regarding this bug? If so >> please say so now. If not I propose to land it in the next day or so. > > This looks all pretty good to me. > I added a few comments to the BZ. All minor stuff except the absence of > insn tests. > > One thing I'd like to bring up is a nightly build with results posted to > the mailing list. We already have enough ports without those and > generally no idea how well those ports work and/or how they brake when > new stuff gets added. We want to be able to observe that. Is there any (free) emulator software for TileGX which could fill that testing gap somehow ? As I knew, there is no free simulator available for TileGX for now. ---- Bye, Roland -- __ . . __ (o.\ \/ /.o) rol...@nr... \__\/\/__/ MPEG specialist, C&&JAVA&&Sun&&Unix programmer /O /==\ O\ TEL +49 641 3992797 (;O/ \/ \O;) |
|
From: Philippe W. <phi...@sk...> - 2015-04-09 19:10:37
|
On Thu, 2015-04-09 at 18:46 +0200, Florian Krohm wrote: > One thing I'd like to bring up is a nightly build with results posted to > the mailing list. We already have enough ports without those and > generally no idea how well those ports work and/or how they brake when > new stuff gets added. We want to be able to observe that. +1. >From my point of view, I would even say addition of a brand new OS or brand new arch implies: * Mandatory requirement: have a nightly build setup * Not mandatory, but close: have an accessible machine by valgrind developers (in gcc compile farm or similarly accessible by valdev). Philippe |
|
From: Julian S. <js...@ac...> - 2015-04-10 11:11:41
|
On 09/04/15 21:11, Philippe Waroquiers wrote: >>From my point of view, I would even say addition of a brand new OS > or brand new arch implies: > * Mandatory requirement: have a nightly build setup > * Not mandatory, but close: have an accessible machine by valgrind > developers (in gcc compile farm or similarly accessible by valdev). Yes, I agree. J |
|
From: Florian K. <fl...@ei...> - 2015-04-09 20:00:43
|
On 09.04.2015 21:11, Philippe Waroquiers wrote: > On Thu, 2015-04-09 at 18:46 +0200, Florian Krohm wrote: > >> One thing I'd like to bring up is a nightly build with results posted to >> the mailing list. We already have enough ports without those and >> generally no idea how well those ports work and/or how they brake when >> new stuff gets added. We want to be able to observe that. > > +1. > >>From my point of view, I would even say addition of a brand new OS > or brand new arch implies: > * Mandatory requirement: have a nightly build setup seconded! > * Not mandatory, but close: have an accessible machine by valgrind > developers (in gcc compile farm or similarly accessible by valdev). > For sure would be nice but probably not realistic. In most companies granting external access to a machine would require an act of Congress or something like that... Florian |
|
From: Philippe W. <phi...@sk...> - 2015-04-09 20:06:08
|
On Thu, 2015-04-09 at 22:00 +0200, Florian Krohm wrote: > > * Not mandatory, but close: have an accessible machine by valgrind > > developers (in gcc compile farm or similarly accessible by valdev). > > > > For sure would be nice but probably not realistic. In most companies > granting external access to a machine would require an act of Congress > or something like that... Yes :). A maybe more realistic approach for such companies is to donate a machine to gcc compile farm (which also avoids the company the administration effort). Philippe |
|
From: Zhigang L. <zl...@ez...> - 2015-04-09 20:40:48
|
Florian an all who concerns, I am in the middle to address your new comments, especially, the format issue of "guest_tilegx_toIR.c", which is based on a generated template file. I have to do it line by line to correct the format, I hope I could finish today, unless you think we could do it after the merge. This was the regression test result I had run with the patches last time. I could improve it further. About half of the failures are due to line # given by stack trace, the line # are within +/- 3 range compared to the expected. Apparently the stack trace code need some enhancement. We could do it later on. ----------------- == 539 tests, 21 stderr failures, 1 stdout failure, 1 stderrB failure, 3 stdoutB failures, 1 post failure == Thanks ZhiGang -----Original Message----- From: Florian Krohm [mailto:fl...@ei...] Sent: Thursday, April 09, 2015 12:47 PM To: js...@ac...; Valgrind Developers Cc: Zhigang Liu Subject: Re: [Valgrind-developers] Linux/TileGX port: last call for comments On 09.04.2015 15:48, Julian Seward wrote: > > Bug 339778 (https://bugs.kde.org/show_bug.cgi?id=339778) contains > patches for a port to TileGx, a 64 bit CPU. There has been quite some > reviewing and re-working of the patches. From my point of view they > are now ready to land. > > Are there any more comments or concerns regarding this bug? If so > please say so now. If not I propose to land it in the next day or so. This looks all pretty good to me. I added a few comments to the BZ. All minor stuff except the absence of insn tests. One thing I'd like to bring up is a nightly build with results posted to the mailing list. We already have enough ports without those and generally no idea how well those ports work and/or how they brake when new stuff gets added. We want to be able to observe that. Florian |
|
From: Zhigang L. <zl...@ez...> - 2015-04-10 02:30:34
|
________________________________________ From: Florian Krohm <fl...@ei...> Sent: Thursday, April 9, 2015 12:46 PM To: js...@ac...; Valgrind Developers Cc: Zhigang Liu Subject: Re: [Valgrind-developers] Linux/TileGX port: last call for comments On 09.04.2015 15:48, Julian Seward wrote: > > Bug 339778 (https://bugs.kde.org/show_bug.cgi?id=339778) contains > patches for a port to TileGx, a 64 bit CPU. There has been quite > some reviewing and re-working of the patches. From my point of > view they are now ready to land. > > Are there any more comments or concerns regarding this bug? If so > please say so now. If not I propose to land it in the next day or so. This looks all pretty good to me. I added a few comments to the BZ. All minor stuff except the absence of insn tests. [ZhiGang] We will add instruction tests separately. It is not ready now. One thing I'd like to bring up is a nightly build with results posted to the mailing list. We already have enough ports without those and generally no idea how well those ports work and/or how they brake when new stuff gets added. We want to be able to observe that. [Zhigang] The regress test shows 20+ failures as post in my previous email today. Most of them are due to the mismatch source file line # given by stack trace. The mismatched line number are very close to expected within +/-3. Apparently, we need improve the stack back trace later. Regarding the maintaining: we can periodically(say once a week) pull the trunk and run on TileGX linux machine and post the results on this mail-list. Florian |
|
From: Philippe W. <phi...@sk...> - 2015-04-10 19:01:16
|
On Fri, 2015-04-10 at 02:30 +0000, Zhigang Liu wrote: > [Zhigang] The regress test shows 20+ failures as post in my previous email today. > Most of them are due to the mismatch source file line # given by stack trace. > The mismatched line number are very close to expected within +/-3. > Apparently, we need improve the stack back trace later. You might compare the stacktraces produced by Valgrind and the stacktraces produced by gdb (using V gdbserver). This might indicate if the problem lies in the Valgrind unwinder or just in the way the compiler generates unwind info. > Regarding the maintaining: we can periodically(say once a week) pull the trunk and > run on TileGX linux machine and post the results on this mail-list. There is a standard script that is used by all nightly builds to build, compare and send the results to the valdev result mailing lists. The idea is to run that nightly (so as to detect problems/regressions) asap. If your company can donate a machine to gcc compile farm, the nightly build can be setup there. Otherwise, a cronjob starting the build in a quiet period is the best thing to do. Philippe |
|
From: Zhigang L. <zl...@ez...> - 2015-04-10 02:34:35
|
________________________________________ From: Philippe Waroquiers <phi...@sk...> Sent: Thursday, April 9, 2015 4:06 PM To: Florian Krohm Cc: js...@ac...; Valgrind Developers; Zhigang Liu Subject: Re: [Valgrind-developers] Linux/TileGX port: last call for comments On Thu, 2015-04-09 at 22:00 +0200, Florian Krohm wrote: > > * Not mandatory, but close: have an accessible machine by valgrind > > developers (in gcc compile farm or similarly accessible by valdev). > > > > For sure would be nice but probably not realistic. In most companies > granting external access to a machine would require an act of Congress > or something like that... Yes :). A maybe more realistic approach for such companies is to donate a machine to gcc compile farm (which also avoids the company the administration effort). I will check if we did that already. Probably not :) Philippe |
|
From: Philippe W. <phi...@sk...> - 2015-04-10 18:41:19
|
On Fri, 2015-04-10 at 02:34 +0000, Zhigang Liu wrote: [about donate a machine to gcc compile farm] > > I will check if we did that already. Probably not :) Would be nice to have such a machine donated. That will for sure help to keep the TileGX Valgrind port in a good state, and also other free software (gcc, gdb, ...) Philippe |