Bug#921274: teeworlds: baseline violation on i386
On Sun, Feb 10, 2019 at 06:35:06PM +, Jordy RUIZ wrote: > Greetings, Hi Jordy, > I'm Dune, coming from upstream. > This should be fixed by > https://github.com/teeworlds/teeworlds/commit/fdc14f07386272c47a95e060643620f537ab9d5e > for bam and https://github.com/teeworlds/teeworlds/pull/2033 for cmake. > Teeworlds will only use the SSE flag if necessary (old gcc compilers, gcc < > 4.10). I hope that does it for you, Adrian. thanks, looks good to me (there was no gcc 4.10 since gcc 5 followed after 4.9, but that has no practical relevance here). > Jordy. cu Adrian -- "Is there not promise of rain?" Ling Tan asked suddenly out of the darkness. There had been need of rain for many days. "Only a promise," Lao Er said. Pearl S. Buck - Dragon Seed
Bug#921274: teeworlds: baseline violation on i386
Greetings, I'm Dune, coming from upstream. This should be fixed by https://github.com/teeworlds/teeworlds/commit/fdc14f07386272c47a95e060643620f537ab9d5e for bam and https://github.com/teeworlds/teeworlds/pull/2033 for cmake. Teeworlds will only use the SSE flag if necessary (old gcc compilers, gcc < 4.10). I hope that does it for you, Adrian. Jordy.
Bug#921274: teeworlds: baseline violation on i386
On Sun, Feb 10, 2019 at 12:33:12PM +0100, Markus Koschany wrote: > Control: severity -1 important > > > Am 10.02.19 um 12:20 schrieb Adrian Bunk: > [...] > > Teeworlds runs fine on some i386 machines. > > > > Individual packages cannot just use non-baseline features like > > SSE or AVX without runtime detection, this results in nothing > > but crashes on hardware officially supported by Debian. > > I disagree with the severity and your handling of this bug report. > > As I wrote, I don't experience any crashes on i386 hardware and upstream > would have certainly done something about it because i386 is officially > supported by them. >... It obviously works if your i386 machine has SSE. I am adding the release team to Cc, asking for a decision whether baseline violations are permitted or considered release critical. > Markus cu Adrian -- "Is there not promise of rain?" Ling Tan asked suddenly out of the darkness. There had been need of rain for many days. "Only a promise," Lao Er said. Pearl S. Buck - Dragon Seed
Bug#921274: teeworlds: baseline violation on i386
Control: severity -1 important Am 10.02.19 um 12:20 schrieb Adrian Bunk: [...] > Teeworlds runs fine on some i386 machines. > > Individual packages cannot just use non-baseline features like > SSE or AVX without runtime detection, this results in nothing > but crashes on hardware officially supported by Debian. I disagree with the severity and your handling of this bug report. As I wrote, I don't experience any crashes on i386 hardware and upstream would have certainly done something about it because i386 is officially supported by them. > >> Your patch also changes >> the upstream build system. Did you talk to upstream about this? What did >> they say? > > I assume the maintainer is best suited for communicaing with upstream. This is a non debian-specific problem and you seem to be the one who disagrees with an upstream decision. You should take this to them, be constructive and then ask for the inclusion in the Debian package before we freeze. Markus signature.asc Description: OpenPGP digital signature
Bug#921274: teeworlds: baseline violation on i386
Control: severity -1 serious Control: tags -1 moreinfo On Sun, Feb 10, 2019 at 12:00:48PM +0100, Markus Koschany wrote: > Control: tags -1 moreinfo > Control: severity -1 important > > On Wed, 6 Feb 2019 22:39:02 +0100 Markus Koschany wrote: > > On Sun, 03 Feb 2019 22:35:22 +0200 Adrian Bunk wrote: > > > Source: teeworlds > > > Version: 0.7.2-2 > > > Severity: serious > > > Tags: patch > > > > > > SSE is not part of the i386 baseline, fix attached. > > > > Could you go into more detail why this is release-critical and what > > issue we are trying to solve? > > > > Markus > > > > Hi, > > since I have not heard back from you, I am going to downgrade the > severity to important. Note that I'm still willing to fix bugs in > teeworlds but I need more information what problem we are trying to > solve. Teeworlds builds and runs fine on i386. >... Teeworlds runs fine on some i386 machines. Individual packages cannot just use non-baseline features like SSE or AVX without runtime detection, this results in nothing but crashes on hardware officially supported by Debian. > Your patch also changes > the upstream build system. Did you talk to upstream about this? What did > they say? I assume the maintainer is best suited for communicaing with upstream. cu Adrian -- "Is there not promise of rain?" Ling Tan asked suddenly out of the darkness. There had been need of rain for many days. "Only a promise," Lao Er said. Pearl S. Buck - Dragon Seed
Bug#921274: teeworlds: baseline violation on i386
Control: tags -1 moreinfo Control: severity -1 important On Wed, 6 Feb 2019 22:39:02 +0100 Markus Koschany wrote: > On Sun, 03 Feb 2019 22:35:22 +0200 Adrian Bunk wrote: > > Source: teeworlds > > Version: 0.7.2-2 > > Severity: serious > > Tags: patch > > > > SSE is not part of the i386 baseline, fix attached. > > Could you go into more detail why this is release-critical and what > issue we are trying to solve? > > Markus > Hi, since I have not heard back from you, I am going to downgrade the severity to important. Note that I'm still willing to fix bugs in teeworlds but I need more information what problem we are trying to solve. Teeworlds builds and runs fine on i386. Your patch also changes the upstream build system. Did you talk to upstream about this? What did they say? signature.asc Description: OpenPGP digital signature
Bug#921274: teeworlds: baseline violation on i386
On Sun, 03 Feb 2019 22:35:22 +0200 Adrian Bunk wrote: > Source: teeworlds > Version: 0.7.2-2 > Severity: serious > Tags: patch > > SSE is not part of the i386 baseline, fix attached. Could you go into more detail why this is release-critical and what issue we are trying to solve? Markus signature.asc Description: OpenPGP digital signature
Bug#921274: teeworlds: baseline violation on i386
Source: teeworlds Version: 0.7.2-2 Severity: serious Tags: patch SSE is not part of the i386 baseline, fix attached. Description: SSE is not part of the i386 baseline -msse2 is not required for _mm_pause(). Author: Adrian Bunk --- teeworlds-0.7.2.orig/bam.lua +++ teeworlds-0.7.2/bam.lua @@ -196,7 +196,6 @@ end function GenerateLinuxSettings(settings, conf, arch, compiler) if arch == "x86" then - settings.cc.flags:Add("-msse2") -- for the _mm_pause call settings.cc.flags:Add("-m32") settings.link.flags:Add("-m32") elseif arch == "x86_64" then