Once upon a time, Florian Weimer <fweimer(a)redhat.com&gt; said: > I ran into this unannounced change: > > https://fedoraproject.org/wiki/Changes/Stop_Building_i686_Kernels I'm assuming it's a work-in-progress proposal? For one thing, it says it is not a system wide change, but (if I'm understanding it correctly) this would be a significant change, as it would remove a supported architecture completely.

On Tuesday, 11 July 2017 at 22:26, Florian Weimer wrote: > I ran into this unannounced change: > > https://fedoraproject.org/wiki/Changes/Stop_Building_i686_Kernels I noticed this is categorized as self-contained, which I think is wrong. I also have hardware that would no longer run Fedora after such change (a netbook with an older Intel Atom CPU which supports SSE2, but is 32bit). Unless the change proponent can provide some numbers suggesting that 32bit users are a tiny minority of our userbase, I'll probably be against such change.

On 11 July 2017 at 16:57, Josh Boyer <jwboyer(a)fedoraproject.org&gt; wrote: > On Tue, Jul 11, 2017 at 4:41 PM, Dominik 'Rathann' Mierzejewski > <dominik(a)greysector.net&gt; wrote: >> On Tuesday, 11 July 2017 at 22:26, Florian Weimer wrote: >>> I ran into this unannounced change: >>> >>> https://fedoraproject.org/wiki/Changes/Stop_Building_i686_Kernels >> >> I noticed this is categorized as self-contained, which I think is wrong. >> >> I also have hardware that would no longer run Fedora after such change >> (a netbook with an older Intel Atom CPU which supports SSE2, but is >> 32bit). Unless the change proponent can provide some numbers suggesting >> that 32bit users are a tiny minority of our userbase, I'll probably >> be against such change. > > Anyone with 32-bit hardware is going to be against this change. It is > a known downside. It also doesn't change the fact that i686 kernels > are in a zombie state, where the kernel team does not actively support > them and the community has not significantly stepped up to do so. > That approach was done quite a while ago, and explicitly communicated. > > The fact that i686 kernels continue to work in general is basically luck. Or that they have been broken at various times and no one noticed.

>The fact that i686 kernels continue to work in general is basically luck. You probably will deny this, but in practice it has been so for many years, because the i686 has dropped out of RHAT's business interest.

> support. As noted in this thread, this isn't just Red Hat -- it is true > of upstream i686 as well. No one is really interested in this. I > guarantee you that if some non-Red Hat person showed up and said "Hey, > I'm here to work on i686 N hours per week", we would say "awesome", not > "Red Hat doesn't care". Would it be possible to make this a Prioritized Bug? It seems to be a classic case of "affects a lot of people, nobody seems to want to take interest".

To give an example outside of the kernel, the installer 'Reclaim Space' function has been broken on i686 for about 10 months, and no-one seems to be lining up to fix that one either: https://bugzilla.redhat.com/show_bug.cgi?id=1375732

On Wed, Jul 12, 2017 at 9:40 PM, Stephen John Smoogen <smooge(a)gmail.com&gt; wrote: > On 12 July 2017 at 16:32, Adam Williamson <adamwill(a)fedoraproject.org&gt; wrote: > >> To give an example outside of the kernel, the installer 'Reclaim Space' >> function has been broken on i686 for about 10 months, and no-one seems >> to be lining up to fix that one either: >> >> https://bugzilla.redhat.com/show_bug.cgi?id=1375732 > > Thanks I have added it to https://fedoraproject.org/wiki/Architectures/x86 Probably useful to setup a x86-32 tracking bug in bugzilla like the other non x86 arches have eg: https://bugzilla.redhat.com/show_bug.cgi?id=245418

It's not a question of a single specific bug, though. It's a question of having someone or a group of someones interesting in the *ongoing requirement* for making sure i686 still works. To give an example outside of the kernel, the installer 'Reclaim Space' function has been broken on i686 for about 10 months, and no-one seems to be lining up to fix that one either:

On Thu, Jul 13, 2017 at 07:47:58AM -0400, Bastien Nocera wrote: > > To give an example outside of the kernel, the installer 'Reclaim Space' > > function has been broken on i686 for about 10 months, and no-one seems > > to be lining up to fix that one either: > To be fair, no-one seems to be lining up to fix anaconda bugs in general... I don't think that's fair at all. There is a whole anaconda team that

very definitely works on bugs. If there's something in particular that needs priortization and you're having trouble getting attention on it, please raise it as a prioritized issue.

> I don't think that's fair at all. There is a whole anaconda team that Case in point. The anaconda team fixes bugs in anaconda. There are few (no?) outside contributors.

On Wed, Jul 12, 2017 at 08:44:02AM -0400, Matthew Miller wrote: > On Wed, Jul 12, 2017 at 01:20:58PM +0200, Ralf Corsepius wrote: >>> The fact that i686 kernels continue to work in general is basically luck. >> You probably will deny this, but in practice it has been so for many >> years, because the i686 has dropped out of RHAT's business interest. > > I don't think this is unreasonable. It is easy for us to support > architectures that a company is paying people to support. It is hard > for us to support architectures that are not getting that that kind of > support. As noted in this thread, this isn't just Red Hat -- it is true > of upstream i686 as well. No one is really interested in this. I > guarantee you that if some non-Red Hat person showed up and said "Hey, > I'm here to work on i686 N hours per week", we would say "awesome", not > "Red Hat doesn't care". Would it be possible to make this a Prioritized Bug? It seems to be a classic case of "affects a lot of people, nobody seems to want to take interest".

On Tue, Jul 11, 2017 at 4:41 PM, Dominik 'Rathann' Mierzejewski <dominik(a)greysector.net&gt; wrote: > On Tuesday, 11 July 2017 at 22:26, Florian Weimer wrote: >> I ran into this unannounced change: >> >> https://fedoraproject.org/wiki/Changes/Stop_Building_i686_Kernels > > I noticed this is categorized as self-contained, which I think is wrong. > > I also have hardware that would no longer run Fedora after such change > (a netbook with an older Intel Atom CPU which supports SSE2, but is > 32bit). Unless the change proponent can provide some numbers suggesting > that 32bit users are a tiny minority of our userbase, I'll probably > be against such change. Anyone with 32-bit hardware is going to be against this change. It is a known downside. It also doesn't change the fact that i686 kernels are in a zombie state, where the kernel team does not actively support them and the community has not significantly stepped up to do so.

On Wed, Jul 12, 2017 at 9:09 AM, Hans de Goede <hdegoede(a)redhat.com&gt; wrote: > Hi, > > On 11-07-17 22:57, Josh Boyer wrote: >> >> On Tue, Jul 11, 2017 at 4:41 PM, Dominik 'Rathann' Mierzejewski >> <dominik(a)greysector.net&gt; wrote: >>> >>> On Tuesday, 11 July 2017 at 22:26, Florian Weimer wrote: >>>> >>>> I ran into this unannounced change: >>>> >>>> https://fedoraproject.org/wiki/Changes/Stop_Building_i686_Kernels >>> >>> >>> I noticed this is categorized as self-contained, which I think is wrong. >>> >>> I also have hardware that would no longer run Fedora after such change >>> (a netbook with an older Intel Atom CPU which supports SSE2, but is >>> 32bit). Unless the change proponent can provide some numbers suggesting >>> that 32bit users are a tiny minority of our userbase, I'll probably >>> be against such change. >> >> >> Anyone with 32-bit hardware is going to be against this change. It is >> a known downside. It also doesn't change the fact that i686 kernels >> are in a zombie state, where the kernel team does not actively support >> them and the community has not significantly stepped up to do so. > > > I still have (some) 32 bit hardware in use and I must say that I was > not aware of this zombie state. i686 kernels have been working fine for > me otherwise I would have likely stepped up to fix things (or if that > was too much work replace my last 32bit hardware), but I may just have > been lucky and never hit a bad kernel. > > If the kernel team wants some specific help with ia32 support then > 2 things need to happen: > > 1) A clear request for help needs to be send > 2) What exactly they need help with needs to be clearly defined https://lists.fedoraproject.org/pipermail/devel/2015-February/208368.html

On Wed, Jul 12, 2017 at 10:14 AM, Hans de Goede <hdegoede(a)redhat.com&gt; wrote: > Hi, > > > On 12-07-17 15:18, Josh Boyer wrote: >> >> On Wed, Jul 12, 2017 at 9:09 AM, Hans de Goede <hdegoede(a)redhat.com&gt; >> wrote: >>> >>> Hi, >>> >>> On 11-07-17 22:57, Josh Boyer wrote: >>>> >>>> >>>> On Tue, Jul 11, 2017 at 4:41 PM, Dominik 'Rathann' Mierzejewski >>>> <dominik(a)greysector.net&gt; wrote: >>>>> >>>>> >>>>> On Tuesday, 11 July 2017 at 22:26, Florian Weimer wrote: >>>>>> >>>>>> >>>>>> I ran into this unannounced change: >>>>>> >>>>>> https://fedoraproject.org/wiki/Changes/Stop_Building_i686_Kernels >>>>> >>>>> >>>>> >>>>> I noticed this is categorized as self-contained, which I think is >>>>> wrong. >>>>> >>>>> I also have hardware that would no longer run Fedora after such change >>>>> (a netbook with an older Intel Atom CPU which supports SSE2, but is >>>>> 32bit). Unless the change proponent can provide some numbers suggesting >>>>> that 32bit users are a tiny minority of our userbase, I'll probably >>>>> be against such change. >>>> >>>> >>>> >>>> Anyone with 32-bit hardware is going to be against this change. It is >>>> a known downside. It also doesn't change the fact that i686 kernels >>>> are in a zombie state, where the kernel team does not actively support >>>> them and the community has not significantly stepped up to do so. >>> >>> >>> >>> I still have (some) 32 bit hardware in use and I must say that I was >>> not aware of this zombie state. i686 kernels have been working fine for >>> me otherwise I would have likely stepped up to fix things (or if that >>> was too much work replace my last 32bit hardware), but I may just have >>> been lucky and never hit a bad kernel. >>> >>> If the kernel team wants some specific help with ia32 support then >>> 2 things need to happen: >>> >>> 1) A clear request for help needs to be send >>> 2) What exactly they need help with needs to be clearly defined >> >> >> https://lists.fedoraproject.org/pipermail/devel/2015-February/208368.html > > > Which is yet another generic, non specific call for help. Which > unsurprisingly (given its unspecificness) did probably not get > a lot of response. > > What would be helpful is a concrete list of things people who > care about i686 can work about. For example an i686 kernel tracker > bug + link to that on the https://fedoraproject.org/wiki/Kernel > page. That's a great idea. Are you volunteering to triage all the bugs and create it and maintain that list? > I believe that something among the lines of: "we need help, but we > are not really specifying what, still please do something" is not > going to get you a lot of help. > > OTOH "here is a prioritized list of TODO items, if all of the > high prio items have not been solved before $date, then we are > going to have to drop ia32 support from F28" OTOH will likely > be much more effective IMHO. This will cut 2 ways: > 1) It will likely get the kernel team more help > 2) If the kernel team does not get help, or not enough, then > you have a strong argument that not enough people care about > ia32 bit support and it should be dropped I have no vested interest in the kernel team any longer other than history and sympathy for the workload they face, but it is 100% acceptable for a team with that much workload to simply say "we are not prioritizing this anymore" irrespective of what other people want. To be honest, they've done much more to keep it building and working than I would have. So while you may want them to define concrete actionable items, I don't find that acceptable. They don't do that for arm, ppc64le, or s390x yet those continue to build and work just fine. People that care about i686 should be able to step in and define those themselves. If they can't then I'd suggest the kernel team just ExcludeArch i686 at the next build break. At least that is an immediate actionable item.

On Wed, Jul 12, 2017 at 12:23 PM, Hans de Goede <hdegoede(a)redhat.com&gt; wrote:

> I find this response not helpful at all, what I see happening here is: > > Person a: We need help > Person b: Ok what can I do > Person a: We need help figure it out yourself > Person b: Huh, so there is nothing to do. Ok then I guess I will go and do > something else. No, you're completely misconstruing the email thread I linked to. We said people interested in helping could do so, and gave a link to some bugs that could be looked at. But the crux of the original post wasn't a cry for help. It was a clarification of where the kernel team would be spending their time. So to phrase it in conversational form it's more like: Person a: We're going to be spending our time looking at x86_64 and fixing issues there. i686 isn't a priority for us. Person b: <silence>

> Feel free to substitute me for person b because any desire I had > to make time to help with this just went out the window. OK. I'm not concerned about that at all.

Hi, On 12-07-17 16:33, Josh Boyer wrote: > On Wed, Jul 12, 2017 at 10:14 AM, Hans de Goede <hdegoede(a)redhat.com&gt; wrote: >> Hi, >> >> >> On 12-07-17 15:18, Josh Boyer wrote: >>> >>> On Wed, Jul 12, 2017 at 9:09 AM, Hans de Goede <hdegoede(a)redhat.com&gt; >>> wrote: >>>> >>>> Hi, >>>> >>>> On 11-07-17 22:57, Josh Boyer wrote: >>>>> >>>>> >>>>> On Tue, Jul 11, 2017 at 4:41 PM, Dominik 'Rathann' Mierzejewski >>>>> <dominik(a)greysector.net&gt; wrote: >>>>>> >>>>>> >>>>>> On Tuesday, 11 July 2017 at 22:26, Florian Weimer wrote: >>>>>>> >>>>>>> >>>>>>> I ran into this unannounced change: >>>>>>> >>>>>>> https://fedoraproject.org/wiki/Changes/Stop_Building_i686_Kernels >>>>>> >>>>>> >>>>>> >>>>>> I noticed this is categorized as self-contained, which I think is >>>>>> wrong. >>>>>> >>>>>> I also have hardware that would no longer run Fedora after such change >>>>>> (a netbook with an older Intel Atom CPU which supports SSE2, but is >>>>>> 32bit). Unless the change proponent can provide some numbers suggesting >>>>>> that 32bit users are a tiny minority of our userbase, I'll probably >>>>>> be against such change. >>>>> >>>>> >>>>> >>>>> Anyone with 32-bit hardware is going to be against this change. It is >>>>> a known downside. It also doesn't change the fact that i686 kernels >>>>> are in a zombie state, where the kernel team does not actively support >>>>> them and the community has not significantly stepped up to do so. >>>> >>>> >>>> >>>> I still have (some) 32 bit hardware in use and I must say that I was >>>> not aware of this zombie state. i686 kernels have been working fine for >>>> me otherwise I would have likely stepped up to fix things (or if that >>>> was too much work replace my last 32bit hardware), but I may just have >>>> been lucky and never hit a bad kernel. >>>> >>>> If the kernel team wants some specific help with ia32 support then >>>> 2 things need to happen: >>>> >>>> 1) A clear request for help needs to be send >>>> 2) What exactly they need help with needs to be clearly defined >>> >>> >>> https://lists.fedoraproject.org/pipermail/devel/2015-February/208368.html >> >> >> Which is yet another generic, non specific call for help. Which >> unsurprisingly (given its unspecificness) did probably not get >> a lot of response. >> >> What would be helpful is a concrete list of things people who >> care about i686 can work about. For example an i686 kernel tracker >> bug + link to that on the https://fedoraproject.org/wiki/Kernel >> page. > > That's a great idea. Are you volunteering to triage all the bugs and > create it and maintain that list? > >> I believe that something among the lines of: "we need help, but we >> are not really specifying what, still please do something" is not >> going to get you a lot of help. >> >> OTOH "here is a prioritized list of TODO items, if all of the >> high prio items have not been solved before $date, then we are >> going to have to drop ia32 support from F28" OTOH will likely >> be much more effective IMHO. This will cut 2 ways: >> 1) It will likely get the kernel team more help >> 2) If the kernel team does not get help, or not enough, then >> you have a strong argument that not enough people care about >> ia32 bit support and it should be dropped > > I have no vested interest in the kernel team any longer other than > history and sympathy for the workload they face, but it is 100% > acceptable for a team with that much workload to simply say "we are > not prioritizing this anymore" irrespective of what other people want. > To be honest, they've done much more to keep it building and working > than I would have. So while you may want them to define concrete > actionable items, I don't find that acceptable. They don't do that > for arm, ppc64le, or s390x yet those continue to build and work just > fine. People that care about i686 should be able to step in and > define those themselves. If they can't then I'd suggest the kernel > team just ExcludeArch i686 at the next build break. At least that is > an immediate actionable item. I find this response not helpful at all, what I see happening here is: Person a: We need help Person b: Ok what can I do Person a: We need help figure it out yourself Person b: Huh, so there is nothing to do. Ok then I guess I will go and do something else. > Feel free to substitute me for person b because any desire I had to make time to help with this just went out the window. More importantly I've the feeling that you can substitute a lot of potential volunteers for person b in the above exchange. To me the we need help or we will just drop support mantra, combined with not willing to put some effort into organizing said help feels a lot like the we need help is just an excuse to just drop ia32 support as that is what you really want to do all along.

Regards, Hans

Hi, On 11-07-17 22:57, Josh Boyer wrote: > > On Tue, Jul 11, 2017 at 4:41 PM, Dominik 'Rathann' Mierzejewski > <dominik(a)greysector.net&gt; wrote: >> >> On Tuesday, 11 July 2017 at 22:26, Florian Weimer wrote: >>> >>> I ran into this unannounced change: >>> >>> https://fedoraproject.org/wiki/Changes/Stop_Building_i686_Kernels >> >> >> I noticed this is categorized as self-contained, which I think is wrong. >> >> I also have hardware that would no longer run Fedora after such change >> (a netbook with an older Intel Atom CPU which supports SSE2, but is >> 32bit). Unless the change proponent can provide some numbers suggesting >> that 32bit users are a tiny minority of our userbase, I'll probably >> be against such change. > > > Anyone with 32-bit hardware is going to be against this change. It is > a known downside. It also doesn't change the fact that i686 kernels > are in a zombie state, where the kernel team does not actively support > them and the community has not significantly stepped up to do so. I still have (some) 32 bit hardware in use and I must say that I was not aware of this zombie state. i686 kernels have been working fine for me otherwise I would have likely stepped up to fix things (or if that was too much work replace my last 32bit hardware), but I may just have been lucky and never hit a bad kernel. If the kernel team wants some specific help with ia32 support then 2 things need to happen: 1) A clear request for help needs to be send

On Wed, Jul 12, 2017 at 7:09 AM, Hans de Goede <hdegoede(a)redhat.com&gt; wrote: > If the kernel team wants some specific help with ia32 support then > 2 things need to happen: > > 1) A clear request for help needs to be send > 2) What exactly they need help with needs to be clearly defined Has happened multiple times over two years. https://lists.fedoraproject.org/pipermail/devel/2015-February/208368. html https://lists.fedoraproject.org/archives/list/devel@lists.fedoraproje ct.org/message/GSIXKLR5RP5OIF6XMDP2WWYOKD4Y5LFM/ https://lists.fedoraproject.org/archives/list/devel@lists.fedoraproje ct.org/message/NJF45R5WFYVDJQJILQEWU2I4ADRYMAFN/ I'm not sure how it could be more clear. Of course, it could have been trolled better by someone, by wiping out the original subject and substituting -- Fedora: "death to i686!" -- > Last as others have said, this is _clearly_ a systemwide change > and the deadline for systemwide change proposals for F27 was July > 4th, where as this page was created July 6th, moreover a change as > big as this really needs to be discussed out in the open long > before > it gets implemented rather then hidden away in a Changes page. I see it both ways. It is self-contained in that it's strictly the kernel without affecting user space at all. But it's system-wide in that it does affect the distribution, it's effectively dropping the hardware support for an entire architecture. But anyway, dropping i686 kernel is not a surprise to me, this has been a rather long leadout train.

I guess if this change goes through I will be looking into how to run a 64-bit kernel and 32-bit userspace. We also still have a few laptops that are running 32-bit Fedora, but they don't have to and they will all be retired within a year anyway.

On Jul 12, 2017, at 15:17, Samuel Sieb <samuel(a)sieb.net&gt; wrote: > On 07/12/2017 12:12 PM, Matthew Smith wrote: >> On Wed, Jul 12, 2017 at 8:03 PM, Samuel Sieb <samuel(a)sieb.net&gt; wrote: >> I guess if this change goes through I >> will be looking into how to run a 64-bit kernel and 32-bit userspace. We >> also still have a few laptops that are running 32-bit Fedora, but they don't >> have to and they will all be retired within a year anyway. > Why do you want to keep the 32-bit userspace? I could be wrong, but since they only have 2GB of RAM, I think that the 32-bit userspace takes less runtime memory. _______________________________________________ devel mailing list -- devel(a)lists.fedoraproject.org To unsubscribe send an email to devel-leave(a)lists.fedoraproject.org

The fact that i686 kernels continue to work in general is basically luck.

Dominik 'Rathann' Mierzejewski wrote: > Considering that SSE2 was introduced by Intel in 2001 and AMD caught up > with it in 2003, I'd be +1 (with my FESCo hat on) to requiring SSE2, > regardless of whether the above change is accepted. If you require SSE2, you limit the usefulness of the i686 kernel to basically just a single generation of CPUs, the next generation introduced x86_64.

On Wednesday, 12 July 2017 at 02:06, Kevin Kofler wrote: > Dominik 'Rathann' Mierzejewski wrote: >> Considering that SSE2 was introduced by Intel in 2001 and AMD caught up >> with it in 2003, I'd be +1 (with my FESCo hat on) to requiring SSE2, >> regardless of whether the above change is accepted. > > If you require SSE2, you limit the usefulness of the i686 kernel to > basically just a single generation of CPUs, the next generation introduced > x86_64. Right. It probably makes sense to abandon i686 kernels altogether, then.

On 07/12/2017 11:57 AM, Dominik 'Rathann' Mierzejewski wrote: > > On Wednesday, 12 July 2017 at 02:06, Kevin Kofler wrote: >> >> Dominik 'Rathann' Mierzejewski wrote: >>> >>> Considering that SSE2 was introduced by Intel in 2001 and AMD caught up >>> with it in 2003, I'd be +1 (with my FESCo hat on) to requiring SSE2, >>> regardless of whether the above change is accepted. >> >> >> If you require SSE2, you limit the usefulness of the i686 kernel to >> basically just a single generation of CPUs, the next generation >> introduced >> x86_64. > > > Right. It probably makes sense to abandon i686 kernels altogether, > then. If you intend to kill Fedora, and furtherly emphasize the impression of Fedora not being community driven distro :(

Dominik 'Rathann' Mierzejewski wrote: > Considering that SSE2 was introduced by Intel in 2001 and AMD caught up > with it in 2003, I'd be +1 (with my FESCo hat on) to requiring SSE2, > regardless of whether the above change is accepted. If you require SSE2, you limit the usefulness of the i686 kernel to basically just a single generation of CPUs, the next generation introduced x86_64.

On Wednesday, 12 July 2017 at 12:04, Christian Dersch wrote: > Well, there is one very important CPU family which came some years > later: Intel Atom. There are still many netbooks out there and the first > Atom generations are 32 bit only afaik. https://en.wikipedia.org/wiki/Intel_Atom The 32 bit-only ones are: Z5xx (April 2008) Z6xx (May 2010) N2xx (June 2008) I still have my N270 netbook, but I guess even more people still have Z6xx-based devices. Still, they're over 7 years old at this point.

On 07/12/2017 12:16 PM, Dominik 'Rathann' Mierzejewski wrote: > I still have my N270 netbook, but I guess even more people still have > Z6xx-based devices. Still, they're over 7 years old at this point. The N270s are still supported by Windows 10. Furthermore: Linux (thus Fedora) has always one of the last resorts for people wanting to escape Windows' and their violent HW obsolesence on older HW. Do you want Fedora to join this choir? I don't.

On Wed, Jul 12, 2017 at 12:30 PM, Ralf Corsepius <rc040203(a)freenet.de&gt; wrote: > On 07/12/2017 12:16 PM, Dominik 'Rathann' Mierzejewski wrote: > >> I still have my N270 netbook, but I guess even more people still have >> Z6xx-based devices. Still, they're over 7 years old at this point. > > > The N270s are still supported by Windows 10. Really? Windows has required uEFI for some time and I never saw a 32 bit n270 with uEFI

On Tue, Jul 11, 2017 at 10:26:03PM +0200, Florian Weimer wrote: > I ran into this unannounced change: > https://fedoraproject.org/wiki/Changes/Stop_Building_i686_Kernels > If this is accepted, all x86 hardware on which Fedora can run will > support SSE2, and we should reflect that in the i686 build flags. > How likely is it that this proposal is accepted? Ideally, we would know > this before the mass rebuild so that we can change the compiler flags in > redhat-rpm-config. Currently i686 users are at about 1/6th of x86_64 users, by mirror checkins. I don't have an easy way of knowing how many of those i686 checkins are old releases -- I'll need to ask Smooge to make a custom report -- but I think it's fair to guess that it's significantly tilted that way. So, taking a SWAG, I'd say maybe 10% of our users would be impacted. That's pretty big, but on the other hand if the cost is disproportionate -- and having heard from the kernel people about this for several years, I think it might be -- it's probably something we should do anyway.

On Tue, Jul 11, 2017 at 10:26:03PM +0200, Florian Weimer wrote: > I ran into this unannounced change: > https://fedoraproject.org/wiki/Changes/Stop_Building_i686_Kernels > If this is accepted, all x86 hardware on which Fedora can run will > support SSE2, and we should reflect that in the i686 build flags. > How likely is it that this proposal is accepted? Ideally, we would know > this before the mass rebuild so that we can change the compiler flags in > redhat-rpm-config. Currently i686 users are at about 1/6th of x86_64 users, by mirror checkins. I don't have an easy way of knowing how many of those i686 checkins are old releases -- I'll need to ask Smooge to make a custom report -- but I think it's fair to guess that it's significantly tilted that way. So, taking a SWAG, I'd say maybe 10% of our users would be impacted. That's pretty big, but on the other hand if the cost is disproportionate -- and having heard from the kernel people about this for several years, I think it might be -- it's probably something we should do anyway. -- Matthew Miller <mattdm(a)fedoraproject.org&gt; Fedora Project Leader _______________________________________________ devel mailing list -- devel(a)lists.fedoraproject.org To unsubscribe send an email to devel-leave(a)lists.fedoraproject.org

Idk, but you know - example - somebody starts with Linux. He won't install it on his home desktop PC or work laptop. He rather find an old dusty laptop from 2005 in his shed and starts to learn there. I think, a lot of people who are potential new users or IT guys, has only access to old HW, nobody else wants. Wasn't that a start for all of us? Or you - as a basic windows user - got brand new high-end laptop for christmas and guess what - installed that weird thing called Linux you never saw before? I'd like to see it supported.

On Tue, Jul 11, 2017 at 10:55:25PM +0200, Michal Schorm wrote: > He won't install it on his home desktop PC or work laptop. He rather find > an old dusty laptop from 2005 in his shed and starts to learn there. If we're being honest, a typical 2005-era laptop is going to yield a rather lousy experience with any modern Linux Desktop. (And an even lousier experience with modern Windows!)

The question is - why - certainly not because CPU got less powerfull that

it's been 10-12 years back - but when gnome clock needs 30MB, and rendering of html pages often takes over 100MB - and nobody cares about any performance regression since new shiny i7 is so fast to 'mask' all programmers faults and 32GB or RAM also needs some usage....

I'm still using some old 32 bit physical servers with Fedora, and they still work well! So I would like to have a 32 bit kernel for some other releases, maybe f28 if possible.

On Tue, Jul 11, 2017 at 10:55:25PM +0200, Michal Schorm wrote: > He won't install it on his home desktop PC or work laptop. He rather find > an old dusty laptop from 2005 in his shed and starts to learn there. If we're being honest, a typical 2005-era laptop is going to yield a rather lousy experience with any modern Linux Desktop. (And an even lousier experience with modern Windows!)

On Tue, Jul 11, 2017 at 10:26:03PM +0200, Florian Weimer wrote: > I ran into this unannounced change: > https://fedoraproject.org/wiki/Changes/Stop_Building_i686_Kernels > If this is accepted, all x86 hardware on which Fedora can run will > support SSE2, and we should reflect that in the i686 build flags. > How likely is it that this proposal is accepted? Ideally, we would know > this before the mass rebuild so that we can change the compiler flags in > redhat-rpm-config. Currently i686 users are at about 1/6th of x86_64 users, by mirror checkins. I don't have an easy way of knowing how many of those i686 checkins are old releases -- I'll need to ask Smooge to make a custom report -- but I think it's fair to guess that it's significantly tilted that way. So, taking a SWAG, I'd say maybe 10% of our users would be impacted. That's pretty big, but on the other hand if the cost is disproportionate -- and having heard from the kernel people about this for several years, I think it might be -- it's probably something we should do anyway.

On Tue, Jul 11, 2017 at 3:43 PM, Matthew Miller <mattdm(a)fedoraproject.org&gt; wrote: > On Tue, Jul 11, 2017 at 10:26:03PM +0200, Florian Weimer wrote: >> I ran into this unannounced change: >> https://fedoraproject.org/wiki/Changes/Stop_Building_i686_Kernels >> If this is accepted, all x86 hardware on which Fedora can run will >> support SSE2, and we should reflect that in the i686 build flags. >> How likely is it that this proposal is accepted? Ideally, we would know >> this before the mass rebuild so that we can change the compiler flags in >> redhat-rpm-config. > > Currently i686 users are at about 1/6th of x86_64 users, by mirror > checkins. I don't have an easy way of knowing how many of those i686 > checkins are old releases -- I'll need to ask Smooge to make a custom > report -- but I think it's fair to guess that it's significantly tilted > that way. So, taking a SWAG, I'd say maybe 10% of our users would be > impacted. That's pretty big, but on the other hand if the cost is > disproportionate -- and having heard from the kernel people about this > for several years, I think it might be -- it's probably something we > should do anyway. > The kernel team quit "supporting" i686 several releases ago, it is down to community support, which is pretty much nonexistent.

Sure, people file bugs, but rarely do people point to or supply patches for those bugs.

Justin

Currently i686 users are at about 1/6th of x86_64 users, by mirror checkins. I don't have an easy way of knowing how many of those i686 checkins are old releases

On Tue, Jul 11, 2017 at 10:26:03PM +0200, Florian Weimer wrote: > I ran into this unannounced change: > https://fedoraproject.org/wiki/Changes/Stop_Building_i686_Kernels > If this is accepted, all x86 hardware on which Fedora can run will > support SSE2, and we should reflect that in the i686 build flags. > How likely is it that this proposal is accepted? Ideally, we would know > this before the mass rebuild so that we can change the compiler flags in > redhat-rpm-config. Currently i686 users are at about 1/6th of x86_64 users, by mirror checkins. I don't have an easy way of knowing how many of those i686 checkins are old releases -- I'll need to ask Smooge to make a custom report -- but I think it's fair to guess that it's significantly tilted that way. So, taking a SWAG, I'd say maybe 10% of our users would be impacted. That's pretty big, but on the other hand if the cost is disproportionate If cost is an issue, consider to drop all these ppc, arm, s370 and mips

On Wed, Jul 12, 2017 at 01:43:53PM +0200, Ralf Corsepius wrote: > If cost is an issue, consider to drop all these ppc, arm, s370 and > mips targets. > > Their user base is like magnitudes smaller than the i686 user base, > while these target are having a significant impact (and thus cost) > on everything in Fedora. Ralf, you know how this works: Fedora is made up of people, and therefore Fedora does what people show up to do. People are showing up to work on those other architectures, even though they are niche, and *no one* is showing up for i686.

https://lists.fedoraproject.org/pipermail/devel/2015-February/208368.html

Which is yet another generic, non specific call for help. Which unsurprisingly (given its unspecificness) did probably not get a lot of response. What would be helpful is a concrete list of things people who care about i686 can work about. For example an i686 kernel tracker bug + link to that on the https://fedoraproject.org/wiki/Kernel page. I believe that something among the lines of: "we need help, but we are not really specifying what, still please do something" is not going to get you a lot of help.

> Currently i686 users are at about 1/6th of x86_64 users, by mirror > checkins. I don't have an easy way of knowing how many of those i686 > checkins are old releases -- I'll need to ask Smooge to make a custom > report -- but I think it's fair to guess that it's significantly tilted > that way. So, taking a SWAG, I'd say maybe 10% of our users would be > impacted. That's pretty big, but on the other hand if the cost is > disproportionate If cost is an issue, consider to drop all these ppc, > arm, s370 and mips Their user base is like magnitudes smaller than the i686 user base, while these target are having a significant impact (and thus cost) on everything in Fedora.

On Tue, Jul 11, 2017 at 10:26:03PM +0200, Florian Weimer wrote: > I ran into this unannounced change: > https://fedoraproject.org/wiki/Changes/Stop_Building_i686_Kernels > If this is accepted, all x86 hardware on which Fedora can run will > support SSE2, and we should reflect that in the i686 build flags. > How likely is it that this proposal is accepted? Ideally, we would know > this before the mass rebuild so that we can change the compiler flags in > redhat-rpm-config. Currently i686 users are at about 1/6th of x86_64 users, by mirror checkins. I don't have an easy way of knowing how many of those i686 checkins are old releases -- I'll need to ask Smooge to make a custom report -- but I think it's fair to guess that it's significantly tilted that way. So, taking a SWAG, I'd say maybe 10% of our users would be impacted. That's pretty big, but on the other hand if the cost is disproportionate -- and having heard from the kernel people about this for several years, I think it might be -- it's probably something we should do anyway.

I ran into this unannounced change: https://fedoraproject.org/wiki/Changes/Stop_Building_i686_Kernels

I ran into this unannounced change: https://fedoraproject.org/wiki/Changes/Stop_Building_i686_Kernels If this is accepted, all x86 hardware on which Fedora can run will support SSE2, and we should reflect that in the i686 build flags.

On 07/11/2017 10:26 PM, Florian Weimer wrote: > I ran into this unannounced change: > > https://fedoraproject.org/wiki/Changes/Stop_Building_i686_Kernels > > If this is accepted, all x86 hardware on which Fedora can run will > support SSE2, and we should reflect that in the i686 build flags. Current GCC does not seem to provide a way to enable SSE2 while preserving stack alignment, but not requiring it for SSE2 access to stack variables. As a result, enabling SSE2 will break tons of third-party software which is not compliant with the increased stack alignment which was introduced as a silent, breaking change to the i386 psABI about 15 years ago. I don't think it's worth fixing GCC because i386 support is legacy only anyway, so I'd suggest to leave things as they are (and even scale back the SSE2 use within glibc).

I ran into this unannounced change: https://fedoraproject.org/wiki/Changes/Stop_Building_i686_Kernels If this is accepted, all x86 hardware on which Fedora can run will support SSE2, and we should reflect that in the i686 build flags.

> https://bugzilla.redhat.com/show_bug.cgi?id=1482798 > (Bug 1482798 - Illegal instruction in SHA1_Update() when used by chronyd) That's definitely a bug for F27.