Re: x86/uv cleanups
On Wed, May 06, 2020 at 04:36:50PM -0500, Russ Anderson wrote: > In addition to Christoph's patches, we will soon be submitting > additional clean up patches. Mike Travis is working on a patch > to remove old SGI UV1 code. Dimitri Sivanich is working on a > sgi_rtc cleanup patch. We are looking at additional cleanup > that should have been done previously. If you plan to submit these very soon I'll happily defer this cleanup series a bit and can resubmit it on top of your changes.
Re: x86/uv cleanups
On Wed, May 06, 2020 at 01:14:40AM +0200, Thomas Gleixner wrote: > Dimitri Sivanich writes: > > Yes, we do see a need to clean up old code where it exists, but we would > > like > > to assume this responsibility ourselves in order to ensure functional > > continuity > > with externally available open-sourced modules that our customers rely on. > > This > > code supports current Superdome Flex systems, as well as forthcoming > > platforms > > based on Intel Cooper Lake and Sapphire Rapids processors. Some cleanup is > > already being included as part of a forthcoming patchset in support of the > > upcoming Sapphire Rapids platform. > > That's a completely bogus argument. > > 1) You could have cleaned up that code long ago > > 2) The rule for code which is unused in the mainline kernel is that it > can be removed during spring cleaning unless there is a compelling > reason. I can't see one here. > > 3) Out of tree modules are not supported independent of the license. > It's not our problem that you have code out of tree which you could > have upstreamed long ago. > > Look at the history of platform/uv code. All we see from you guys is: > > - Drops of adding support for new platforms > > - Occasional fixes when we managed to break UV or you found some bug in >your precious code. > > > Nacked-by: Dimitri Sivanich > > I'm happy to add that tag when merging this cleanup. > > Thanks, > > tglx Thomas (and all) Sorry. We do need to do better. In addition to Christoph's patches, we will soon be submitting additional clean up patches. Mike Travis is working on a patch to remove old SGI UV1 code. Dimitri Sivanich is working on a sgi_rtc cleanup patch. We are looking at additional cleanup that should have been done previously. Steve Wahl will be involved on an ongoing basis, so you will see more from us. Thanks. -- Russ Anderson, SuperDome Flex Linux Kernel Group Manager HPE - Hewlett Packard Enterprise (formerly SGI) r...@hpe.com
Re: x86/uv cleanups
Dimitri Sivanich writes: > Yes, we do see a need to clean up old code where it exists, but we would like > to assume this responsibility ourselves in order to ensure functional > continuity > with externally available open-sourced modules that our customers rely on. > This > code supports current Superdome Flex systems, as well as forthcoming platforms > based on Intel Cooper Lake and Sapphire Rapids processors. Some cleanup is > already being included as part of a forthcoming patchset in support of the > upcoming Sapphire Rapids platform. That's a completely bogus argument. 1) You could have cleaned up that code long ago 2) The rule for code which is unused in the mainline kernel is that it can be removed during spring cleaning unless there is a compelling reason. I can't see one here. 3) Out of tree modules are not supported independent of the license. It's not our problem that you have code out of tree which you could have upstreamed long ago. Look at the history of platform/uv code. All we see from you guys is: - Drops of adding support for new platforms - Occasional fixes when we managed to break UV or you found some bug in your precious code. > Nacked-by: Dimitri Sivanich I'm happy to add that tag when merging this cleanup. Thanks, tglx
Re: x86/uv cleanups
> On May 5, 2020, at 9:25 AM, Dimitri Sivanich wrote: > > Yes, we do see a need to clean up old code where it exists, but we would like > to assume this responsibility ourselves in order to ensure functional > continuity > with externally available open-sourced modules that our customers rely on. > This > code supports current Superdome Flex systems, as well as forthcoming platforms > based on Intel Cooper Lake and Sapphire Rapids processors. Some cleanup is > already being included as part of a forthcoming patchset in support of the > upcoming Sapphire Rapids platform. > What is “functional continuity”? How does unused code support anything? > Nacked-by: Dimitri Sivanich > >> On Mon, May 04, 2020 at 07:15:16PM +0200, Christoph Hellwig wrote: >> Hi x86 maintainers, >> >> this series removes various exports and sniplets of dead code >> from the x86/uv code. >> >> Diffstat: >> >> 8 files changed, 21 insertions(+), 111 deletions(-)
Re: x86/uv cleanups
Yes, we do see a need to clean up old code where it exists, but we would like to assume this responsibility ourselves in order to ensure functional continuity with externally available open-sourced modules that our customers rely on. This code supports current Superdome Flex systems, as well as forthcoming platforms based on Intel Cooper Lake and Sapphire Rapids processors. Some cleanup is already being included as part of a forthcoming patchset in support of the upcoming Sapphire Rapids platform. Nacked-by: Dimitri Sivanich On Mon, May 04, 2020 at 07:15:16PM +0200, Christoph Hellwig wrote: > Hi x86 maintainers, > > this series removes various exports and sniplets of dead code > from the x86/uv code. > > Diffstat: > > 8 files changed, 21 insertions(+), 111 deletions(-)
x86/uv cleanups
Hi x86 maintainers, this series removes various exports and sniplets of dead code from the x86/uv code. Diffstat: 8 files changed, 21 insertions(+), 111 deletions(-)