[opensuse-kernel] revert to 4K pagesize for ppc64
Neither firewire nor nouveau seem to support 64K pagesize. Any objection to revert to 4K pages for ppc64/default? ppc64le can remain 64K. Olaf -- To unsubscribe, e-mail: opensuse-kernel+unsubscribe@opensuse.org To contact the owner, e-mail: opensuse-kernel+owner@opensuse.org
On Sat, Nov 21, Olaf Hering wrote:
Neither firewire nor nouveau seem to support 64K pagesize. Any objection to revert to 4K pages for ppc64/default? ppc64le can remain 64K.
Appearently no objections. I will send a merge request later this week. Olaf -- To unsubscribe, e-mail: opensuse-kernel+unsubscribe@opensuse.org To contact the owner, e-mail: opensuse-kernel+owner@opensuse.org
On Mon, Nov 30, 2015 at 8:38 AM, Olaf Hering <olaf@aepfle.de> wrote:
On Sat, Nov 21, Olaf Hering wrote:
Neither firewire nor nouveau seem to support 64K pagesize. Any objection to revert to 4K pages for ppc64/default? ppc64le can remain 64K.
Appearently no objections. I will send a merge request later this week. Can we do that through flavor? pmac or 4k?
I'm still using bigendian as build worker on power7 machines. And I'd like to keep it unchanged. Dinar,
Olaf -- To unsubscribe, e-mail: opensuse-kernel+unsubscribe@opensuse.org To contact the owner, e-mail: opensuse-kernel+owner@opensuse.org
-- To unsubscribe, e-mail: opensuse-kernel+unsubscribe@opensuse.org To contact the owner, e-mail: opensuse-kernel+owner@opensuse.org
On Mon, Nov 30, Dinar Valeev wrote:
On Mon, Nov 30, 2015 at 8:38 AM, Olaf Hering <olaf@aepfle.de> wrote:
On Sat, Nov 21, Olaf Hering wrote:
Neither firewire nor nouveau seem to support 64K pagesize. Any objection to revert to 4K pages for ppc64/default? ppc64le can remain 64K.
Appearently no objections. I will send a merge request later this week. Can we do that through flavor? pmac or 4k?
Unlikely, its a compiletime setting.
I'm still using bigendian as build worker on power7 machines. And I'd like to keep it unchanged.
How does 64k vs. 4k make a difference? Does the setup use kvm, and if so is kvm unable to run a guest with different pagesize? Olaf -- To unsubscribe, e-mail: opensuse-kernel+unsubscribe@opensuse.org To contact the owner, e-mail: opensuse-kernel+owner@opensuse.org
-----BEGIN PGP SIGNED MESSAGE----- Hash: SHA1 On 11/30/15 9:10 AM, Olaf Hering wrote:
On Mon, Nov 30, Dinar Valeev wrote:
On Mon, Nov 30, 2015 at 8:38 AM, Olaf Hering <olaf@aepfle.de> wrote:
On Sat, Nov 21, Olaf Hering wrote:
Neither firewire nor nouveau seem to support 64K pagesize. Any objection to revert to 4K pages for ppc64/default? ppc64le can remain 64K.
Appearently no objections. I will send a merge request later this week. Can we do that through flavor? pmac or 4k?
Unlikely, its a compiletime setting.
Which is exactly why we have flavors. If it were runtime settable, we could just have a sysctl for it.
I'm still using bigendian as build worker on power7 machines. And I'd like to keep it unchanged.
How does 64k vs. 4k make a difference? Does the setup use kvm, and if so is kvm unable to run a guest with different pagesize?
It's a significant divergence from how we've shipped ppc64 kernels for ages. Two broken drivers on one piece of (old) hardware doesn't justify changing that, IMO. - -Jeff - -- Jeff Mahoney SUSE Labs -----BEGIN PGP SIGNATURE----- Version: GnuPG/MacGPG2 v2.0.19 (Darwin) Comment: GPGTools - http://gpgtools.org iQIcBAEBAgAGBQJWXbA0AAoJEB57S2MheeWylv0P/iUPL5N1V5rfK+2eKgyP/ktk Me0UL7Pkp6EQ1I3cf5Wx9tC1A84MRM7PtLAbf6titr4dmd32PxHfRayF8h674/L5 xd2mqgvJmiNBtKU30uEYb4+8S1eu9JtvK9hTZO580ps1IE9lx0BPPooBfxdS284/ 1FIPZRGYrjNmYKABif2CcYJcQCeHSZ1CgbvL25QNBoWLTs5xzOqHtM5Y+m2RgCJw zm1B+5fphLDiBSfsYZm4/56e3IptQqLU5BbUmwsery0ljyi7eITGNVSXl02KF9Kp 24zN4heL/3Kp9Tj7ssoBjf9ee7fhLym1KlimDu70iMDx+dIYCHodIkVeWDCcchLB d16QqQBe2JSJKQ+xwfXY2aD1oSTh3SBdyvMtQtGXoDJbYBaEZ15cRrOJIc/QhSg4 ywPLaH5H0JotpPXSD1kbkkCmcFLIlqxaWzu6aFTtjglAAWtacVx3Rae01d0Y5CYn s5/e849+nOFmHIQEEsQK4u8NptToANxrHr7zOuMSrLCT5rmZTxdG3QTgvnu0Dmff 6vtBqq27lXuqGhH2fyOjjQIxrDTsIjn6b1ltURxdpyRg/Vjmgug9Lpx/oQ1cdpCp uzBVpilUzwwpZg4VTG730Dn5JfO9trQvZcPfbsozZcqnAaY7wJ9V8F0rDTF6lr5W SystsuoZP6wjonCpK11t =qBCJ -----END PGP SIGNATURE----- -- To unsubscribe, e-mail: opensuse-kernel+unsubscribe@opensuse.org To contact the owner, e-mail: opensuse-kernel+owner@opensuse.org
On Tue, Dec 01, Jeff Mahoney wrote:
Which is exactly why we have flavors. If it were runtime settable, we could just have a sysctl for it.
Are you suggesting a ppc64/4k or /64k flavour?
It's a significant divergence from how we've shipped ppc64 kernels for ages. Two broken drivers on one piece of (old) hardware doesn't justify changing that, IMO.
Who is the consumer of Tumbleweed anyway? Just me, or this there someone else lurking? Olaf -- To unsubscribe, e-mail: opensuse-kernel+unsubscribe@opensuse.org To contact the owner, e-mail: opensuse-kernel+owner@opensuse.org
-----BEGIN PGP SIGNED MESSAGE----- Hash: SHA1 On 12/1/15 9:45 AM, Olaf Hering wrote:
On Tue, Dec 01, Jeff Mahoney wrote:
Which is exactly why we have flavors. If it were runtime settable, we could just have a sysctl for it.
Are you suggesting a ppc64/4k or /64k flavour?
I'm suggesting a 4k flavor. 64k is the default and should stay that way .
It's a significant divergence from how we've shipped ppc64 kernels for ages. Two broken drivers on one piece of (old) hardware doesn't justify changing that, IMO.
Who is the consumer of Tumbleweed anyway? Just me, or this there someone else lurking?
Dinar already popped up objecting to the change. Torsten agreed with my concerns when I mentioned it on IRC. Do we have numbers anywhere for downloads of ppc64? I don't think it's reasonable to assume that every ppc64be user is lurking on opensuse-kernel. - -Jeff - -- Jeff Mahoney SUSE Labs -----BEGIN PGP SIGNATURE----- Version: GnuPG/MacGPG2 v2.0.19 (Darwin) Comment: GPGTools - http://gpgtools.org iQIcBAEBAgAGBQJWXbP3AAoJEB57S2MheeWyAfQQAIggwvIH4+zHyWFkQpN3u86f mbZx2f/2MGq7+ZogbAkSYAUf52PhKQjKqLqwmu7yZFW9YiF7ATGGSG7AXUOmcOaQ gss84bIOWCdUSTrrvq21A3XA1KxxPHGq0te5sRa8u3QnHJtgni3Sg/JVKLDOF/LM gnZKU8nC3M4n9431B9Oynb1I7kWHI5HyCp+huRaoeRCgunRcndXBf1bJLwxFW7IY 7OAlGo7F4BqBtubscZQOZp9CcOXsNwlWdlMMNOQBugEkM4oXsDvo/u1KacueJDye Lxv4xNgeJuY4xGl0QLz/9VZbrlPw9OW6PgUq5icASr54RPeKmeYih8o4fifkx5+k NxPiF0YP/QeMkizd0YkktQA61y+JRTXW82vHr4kKTvN2sNgkIygxUu4vfRh97R+p Q2P6pLXDZmfAab70WCcDv7abK1TActH1gsMxmza8A4AbV4XasS4emX+MINrQlYnR 9VVrRtJEW5Mo5NkJA0lB/1BvXi3zdBXZyYNyxUC+DrJj1/TaD9KCwGbSzuqUQVsM 4T4BcYRGsjtSKNwdiV/J6rIQdlHpmV0wDD7ELEwhSE6zRjSNlBjMT3bxl1/ODjiK Br5Sw8UPQ1Dfb7REe3iieHKW4CTh2taE5Mmz/MX6npFVZGUEd3QeropH1mepEDvG F5UAuq3Dg6XmgmSG6pK6 =E6nn -----END PGP SIGNATURE----- -- To unsubscribe, e-mail: opensuse-kernel+unsubscribe@opensuse.org To contact the owner, e-mail: opensuse-kernel+owner@opensuse.org
On Tue, Dec 1, 2015 at 3:51 PM, Jeff Mahoney <jeffm@suse.com> wrote:
-----BEGIN PGP SIGNED MESSAGE----- Hash: SHA1
On 12/1/15 9:45 AM, Olaf Hering wrote:
On Tue, Dec 01, Jeff Mahoney wrote:
Which is exactly why we have flavors. If it were runtime settable, we could just have a sysctl for it.
Are you suggesting a ppc64/4k or /64k flavour?
I'm suggesting a 4k flavor. 64k is the default and should stay that way .
It's a significant divergence from how we've shipped ppc64 kernels for ages. Two broken drivers on one piece of (old) hardware doesn't justify changing that, IMO.
Who is the consumer of Tumbleweed anyway? Just me, or this there someone else lurking?
Dinar already popped up objecting to the change. Torsten agreed with my concerns when I mentioned it on IRC. Do we have numbers anywhere for downloads of ppc64? I don't think it's reasonable to assume that every ppc64be user is lurking on opensuse-kernel. We don't have much users, sure. But we have some on old p5-p7 servers.
- -Jeff
- -- Jeff Mahoney SUSE Labs -----BEGIN PGP SIGNATURE----- Version: GnuPG/MacGPG2 v2.0.19 (Darwin) Comment: GPGTools - http://gpgtools.org
iQIcBAEBAgAGBQJWXbP3AAoJEB57S2MheeWyAfQQAIggwvIH4+zHyWFkQpN3u86f mbZx2f/2MGq7+ZogbAkSYAUf52PhKQjKqLqwmu7yZFW9YiF7ATGGSG7AXUOmcOaQ gss84bIOWCdUSTrrvq21A3XA1KxxPHGq0te5sRa8u3QnHJtgni3Sg/JVKLDOF/LM gnZKU8nC3M4n9431B9Oynb1I7kWHI5HyCp+huRaoeRCgunRcndXBf1bJLwxFW7IY 7OAlGo7F4BqBtubscZQOZp9CcOXsNwlWdlMMNOQBugEkM4oXsDvo/u1KacueJDye Lxv4xNgeJuY4xGl0QLz/9VZbrlPw9OW6PgUq5icASr54RPeKmeYih8o4fifkx5+k NxPiF0YP/QeMkizd0YkktQA61y+JRTXW82vHr4kKTvN2sNgkIygxUu4vfRh97R+p Q2P6pLXDZmfAab70WCcDv7abK1TActH1gsMxmza8A4AbV4XasS4emX+MINrQlYnR 9VVrRtJEW5Mo5NkJA0lB/1BvXi3zdBXZyYNyxUC+DrJj1/TaD9KCwGbSzuqUQVsM 4T4BcYRGsjtSKNwdiV/J6rIQdlHpmV0wDD7ELEwhSE6zRjSNlBjMT3bxl1/ODjiK Br5Sw8UPQ1Dfb7REe3iieHKW4CTh2taE5Mmz/MX6npFVZGUEd3QeropH1mepEDvG F5UAuq3Dg6XmgmSG6pK6 =E6nn -----END PGP SIGNATURE----- -- To unsubscribe, e-mail: opensuse-kernel+unsubscribe@opensuse.org To contact the owner, e-mail: opensuse-kernel+owner@opensuse.org
-- To unsubscribe, e-mail: opensuse-kernel+unsubscribe@opensuse.org To contact the owner, e-mail: opensuse-kernel+owner@opensuse.org
On Tue, Dec 01, Dinar Valeev wrote:
We don't have much users, sure. But we have some on old p5-p7 servers.
And, will they break with 4k PAGE_SIZE? This question was never answered. 64k was introduced for SLE11. For TW I'm sure there will be no regression, modulo the answer to the question above. Also, adding yet another binary and wasting resources for it can hardly be the answer. Olaf -- To unsubscribe, e-mail: opensuse-kernel+unsubscribe@opensuse.org To contact the owner, e-mail: opensuse-kernel+owner@opensuse.org
-----BEGIN PGP SIGNED MESSAGE----- Hash: SHA1 On 12/1/15 9:58 AM, Olaf Hering wrote:
On Tue, Dec 01, Dinar Valeev wrote:
We don't have much users, sure. But we have some on old p5-p7 servers.
And, will they break with 4k PAGE_SIZE? This question was never answered.
64k was introduced for SLE11. For TW I'm sure there will be no regression, modulo the answer to the question above. Also, adding yet another binary and wasting resources for it can hardly be the answer.
It's not. Fixing the broken drivers is the answer. Anything else is a workaround, and I'm not willing to sacrifice performance on non-Apple hardware to work around broken drivers. - -Jeff - -- Jeff Mahoney SUSE Labs -----BEGIN PGP SIGNATURE----- Version: GnuPG/MacGPG2 v2.0.19 (Darwin) Comment: GPGTools - http://gpgtools.org iQIcBAEBAgAGBQJWXbf3AAoJEB57S2MheeWyQEUQAIshJQkYmaKLFdA7n5k7YEDo X8R/FF3njaU5ni+7WjsOqUbF7BK0KFshGJ+eUDyb/xNy18eGB+44JiucYxNZ7Zjd fsP/K6eSPpcDOqqxaNXV/WSMqbheoOuOy4VsyK9hZgJUnCd22nBTBQiUIjJB4xWG LthoPFAlepTZxLC4Xca2ea5NVl3IMeKjt/pCpTsotr/UJJssWrN/41oic9Bn8nsW LfmPkG54Jz8S95b6PZ5xjF2aXd0WuhQpQccacTmV2uBCy1IH+SJANqSKxIoiu9OY Knp4R0hPfHzpV8hizhUNrZ18XuyKP8UD0wtoEKl0SQCwVj+LhCaduCoaerGHudQ2 nOqpJWlFCHYRcJbgTNbcFl7TTuRM2zwm78rzv+n5mGVnIKOn3bZy99/vbe5chI5P Naiz0U7RHkfSeX/OF25HxSLrzbNawi6GbK9TiYU/5pZUG+/+9HW53W+aW1PLIPUz 56yzz4E8AeFgmffEBh7mbWjN+l+Ilwo/axJPviz0oQvjk5TceziCeQJJzUnis3xK 0XRXLG9lrSXkAfV2MFWbdxF3UucJ5Wz21y/sOTzRK/tLOkiMkVk8fysPovh6IBJ9 b4ts6KrR1fQ4HzfyqTrDQ21DJm+fRr1vo6WTvKdH/G1jALBsi0a3ce2azanTlUJW oxsareQEymgNccRAb5tf =tPTk -----END PGP SIGNATURE----- -- To unsubscribe, e-mail: opensuse-kernel+unsubscribe@opensuse.org To contact the owner, e-mail: opensuse-kernel+owner@opensuse.org
On Tue, Dec 01, 2015 at 09:51:35AM -0500, Jeff Mahoney wrote:
Dinar already popped up objecting to the change. Torsten agreed with my concerns when I mentioned it on IRC.
In fact I wondered whether there is a fundamental, conceptual problem with those 2 drivers and 64k page size. Before we discuss here for ages: how much work is it to make those 2 drivers a bit more flexible? Wouldn't that be a big gain for everyone? Torsten -- To unsubscribe, e-mail: opensuse-kernel+unsubscribe@opensuse.org To contact the owner, e-mail: opensuse-kernel+owner@opensuse.org
Am 01.12.2015 um 15:45 schrieb Olaf Hering:
On Tue, Dec 01, Jeff Mahoney wrote:
Which is exactly why we have flavors. If it were runtime settable, we could just have a sysctl for it.
Are you suggesting a ppc64/4k or /64k flavour?
It's a significant divergence from how we've shipped ppc64 kernels for ages. Two broken drivers on one piece of (old) hardware doesn't justify changing that, IMO.
Who is the consumer of Tumbleweed anyway? Just me, or this there someone else lurking?
Me. I had reverted to a 13.2 kernel because of some other driver issue (sr / ata). Re-trying I am still seeing errors but it boots now. Regards, Andreas -- SUSE Linux GmbH, Maxfeldstr. 5, 90409 Nürnberg, Germany GF: Felix Imendörffer, Jane Smithard, Graham Norton; HRB 21284 (AG Nürnberg) -- To unsubscribe, e-mail: opensuse-kernel+unsubscribe@opensuse.org To contact the owner, e-mail: opensuse-kernel+owner@opensuse.org
On Mon, Nov 30, 2015 at 3:10 PM, Olaf Hering <olaf@aepfle.de> wrote:
On Mon, Nov 30, Dinar Valeev wrote:
On Mon, Nov 30, 2015 at 8:38 AM, Olaf Hering <olaf@aepfle.de> wrote:
On Sat, Nov 21, Olaf Hering wrote:
Neither firewire nor nouveau seem to support 64K pagesize. Any objection to revert to 4K pages for ppc64/default? ppc64le can remain 64K.
Appearently no objections. I will send a merge request later this week. Can we do that through flavor? pmac or 4k?
Unlikely, its a compiletime setting.
I'm still using bigendian as build worker on power7 machines. And I'd like to keep it unchanged.
How does 64k vs. 4k make a difference? Does the setup use kvm, and if so is kvm unable to run a guest with different pagesize? Keeping two platforms BE and LE the same as much as possible. We don't do stagings for BE for example, LE only. So more it differs, more changes something broken will sneak in to Tumbleweed
Olaf -- To unsubscribe, e-mail: opensuse-kernel+unsubscribe@opensuse.org To contact the owner, e-mail: opensuse-kernel+owner@opensuse.org
On Mon, Nov 30, 2015 at 3:10 PM, Olaf Hering <olaf@aepfle.de> wrote:
On Mon, Nov 30, Dinar Valeev wrote:
On Mon, Nov 30, 2015 at 8:38 AM, Olaf Hering <olaf@aepfle.de> wrote:
On Sat, Nov 21, Olaf Hering wrote:
Neither firewire nor nouveau seem to support 64K pagesize. Any objection to revert to 4K pages for ppc64/default? ppc64le can remain 64K.
Appearently no objections. I will send a merge request later this week. Can we do that through flavor? pmac or 4k?
Unlikely, its a compiletime setting.
I'm still using bigendian as build worker on power7 machines. And I'd like to keep it unchanged.
How does 64k vs. 4k make a difference? Does the setup use kvm, and if so is kvm unable to run a guest with different pagesize? Keeping two platforms BE and LE the same as much as possible. We don't do stagings for BE for example, LE only. So more it differs, more changes something broken will sneak in to Tumbleweed
Olaf -- To unsubscribe, e-mail: opensuse-kernel+unsubscribe@opensuse.org To contact the owner, e-mail: opensuse-kernel+owner@opensuse.org
On Tue, Dec 01, Dinar Valeev wrote:
Keeping two platforms BE and LE the same as much as possible. We don't do stagings for BE for example, LE only. So more it differs, more changes something broken will sneak in to Tumbleweed
Depending on the view, TW is already broken, or CHRP-only anyway. Olaf -- To unsubscribe, e-mail: opensuse-kernel+unsubscribe@opensuse.org To contact the owner, e-mail: opensuse-kernel+owner@opensuse.org
On Mon, Nov 30, 2015 at 8:38 AM, Olaf Hering <olaf@aepfle.de> wrote:
On Sat, Nov 21, Olaf Hering wrote:
Neither firewire nor nouveau seem to support 64K pagesize. Any objection to revert to 4K pages for ppc64/default? ppc64le can remain 64K.
Appearently no objections. I will send a merge request later this week. Can we do that through flavor? pmac or 4k?
I'm still using bigendian as build worker on power7 machines. And I'd like to keep it unchanged.
Olaf -- To unsubscribe, e-mail: opensuse-kernel+unsubscribe@opensuse.org To contact the owner, e-mail: opensuse-kernel+owner@opensuse.org
-- To unsubscribe, e-mail: opensuse-kernel+unsubscribe@opensuse.org To contact the owner, e-mail: opensuse-kernel+owner@opensuse.org
On Mon, Nov 30, 2015 at 8:38 AM, Olaf Hering <olaf@aepfle.de> wrote:
On Sat, Nov 21, Olaf Hering wrote:
Neither firewire nor nouveau seem to support 64K pagesize. Any objection to revert to 4K pages for ppc64/default? ppc64le can remain 64K.
Appearently no objections. I will send a merge request later this week. Can we do that through flavor? pmac or 4k?
I'm still using bigendian as build worker on power7 machines. And I'd like to keep it unchanged.
Olaf -- To unsubscribe, e-mail: opensuse-kernel+unsubscribe@opensuse.org To contact the owner, e-mail: opensuse-kernel+owner@opensuse.org
-- To unsubscribe, e-mail: opensuse-kernel+unsubscribe@opensuse.org To contact the owner, e-mail: opensuse-kernel+owner@opensuse.org
participants (6)
-
Andreas Färber -
Dinar Valeev -
Dinar Valeev -
Jeff Mahoney -
Olaf Hering -
Torsten Duwe