Re: [Qemu-devel] [PATCH v4 0/3] AHCI migration
Am 18.01.2013 11:28, schrieb Stefan Hajnoczi: On Thu, Jan 17, 2013 at 11:01:52AM +0100, Kevin Wolf wrote: Let's get Jason's patches merged while they still apply. I addressed the review comments (mostly my own) that came up during the v3 review, otherwise this is unchanged. Please note that in my tests it didn't work entirely reliably and I saw guest lockups and kernel crashes in like one of ten cases. I confirmed that the same kind of bugs occurs with v3 of the series, so my changes are likely innocent. Someone will have to debug this some more, but what I did took about the time that I'm willing to spend on it right now. It makes sense to merge these patches to avoid bitrot, but should we keep unmigratable = 1 so that users aren't led to believe migration works? Developers willing to tackle the remaining problem can always comment out the unmigratable flag during testing. But I think users shouldn't be exposed to something unreliable. Makes sense. Added unmigratable = 1 back and applied all to the block branch. Kevin
Re: [Qemu-devel] [PATCH v4 0/3] AHCI migration
On Thu, Jan 17, 2013 at 11:01:52AM +0100, Kevin Wolf wrote: Let's get Jason's patches merged while they still apply. I addressed the review comments (mostly my own) that came up during the v3 review, otherwise this is unchanged. Please note that in my tests it didn't work entirely reliably and I saw guest lockups and kernel crashes in like one of ten cases. I confirmed that the same kind of bugs occurs with v3 of the series, so my changes are likely innocent. Someone will have to debug this some more, but what I did took about the time that I'm willing to spend on it right now. It makes sense to merge these patches to avoid bitrot, but should we keep unmigratable = 1 so that users aren't led to believe migration works? Developers willing to tackle the remaining problem can always comment out the unmigratable flag during testing. But I think users shouldn't be exposed to something unreliable. Stefan
Re: [Qemu-devel] [PATCH v4 0/3] AHCI migration
Stefan Hajnoczi stefa...@redhat.com wrote: On Thu, Jan 17, 2013 at 11:01:52AM +0100, Kevin Wolf wrote: Let's get Jason's patches merged while they still apply. I addressed the review comments (mostly my own) that came up during the v3 review, otherwise this is unchanged. Please note that in my tests it didn't work entirely reliably and I saw guest lockups and kernel crashes in like one of ten cases. I confirmed that the same kind of bugs occurs with v3 of the series, so my changes are likely innocent. Someone will have to debug this some more, but what I did took about the time that I'm willing to spend on it right now. It makes sense to merge these patches to avoid bitrot, but should we keep unmigratable = 1 so that users aren't led to believe migration works? Developers willing to tackle the remaining problem can always comment out the unmigratable flag during testing. But I think users shouldn't be exposed to something unreliable. Agreed with that. Later, Juan.
[Qemu-devel] [PATCH v4 0/3] AHCI migration
Let's get Jason's patches merged while they still apply. I addressed the review comments (mostly my own) that came up during the v3 review, otherwise this is unchanged. Please note that in my tests it didn't work entirely reliably and I saw guest lockups and kernel crashes in like one of ten cases. I confirmed that the same kind of bugs occurs with v3 of the series, so my changes are likely innocent. Someone will have to debug this some more, but what I did took about the time that I'm willing to spend on it right now. Jason Baron (2): ahci: Remove unused AHCIDevice fields ahci: Add migration support Kevin Wolf (1): ahci: Change data types in preparation for migration hw/ide/ahci.c | 97 ++-- hw/ide/ahci.h | 20 --- hw/ide/ich.c | 13 +-- 3 files changed, 109 insertions(+), 21 deletions(-) -- 1.7.6.5