| Message ID | 1406290560-13561-1-git-send-email-pbonzini@redhat.com |
|---|---|
| State | New |
| Headers | show |
Il 25/07/2014 14:15, Paolo Bonzini ha scritto: > The following changes since commit f368c33d5ab09dd5656924185cd975b11838cd25: > > Update version for v2.1.0-rc3 release (2014-07-22 18:17:03 +0100) > > are available in the git repository at: > > git://github.com/bonzini/qemu.git tags/for-upstream > > for you to fetch changes up to d975e28437377bc9a65fb5f4f9486a74c7a124cc: > > qemu-char: ignore flow control if a PTY's slave is not connected (2014-07-24 18:31:50 +0200) > > ---------------------------------------------------------------- > Here are my patches for the last 2.1 rc. > > ---------------------------------------------------------------- > Paolo Bonzini (3): > acpi-dsdt: procedurally generate _PRT > pc: hack for migration compatibility from QEMU 2.0 > qemu-char: ignore flow control if a PTY's slave is not connected Since Igor hasn't sent his patches, and I'm leaving the office, I pushed this to git://github.com/bonzini/qemu.git tags/for-upstream-full I don't know about tests/acpi-test-data/pc. It makes sense that this patch should modify something there, but: * "make check" passes * the test warns even before patch 1, for both the DSDT (modified by the patch) and SSDT (which this series doesn't touch at all) * I cannot get it to pass, except by blindly copying the "actual" output on the "expected" files * mst is on vacation and Marcel is off on Fridays Based on my understanding of the problem, it is not possible to fix the bug without hacks like this one, and even reverting all patches in this area would be more risky. Paolo
On 25 July 2014 15:22, Paolo Bonzini <pbonzini@redhat.com> wrote: > Since Igor hasn't sent his patches, and I'm leaving the office, I pushed > this to > > git://github.com/bonzini/qemu.git tags/for-upstream-full > > I don't know about tests/acpi-test-data/pc. It makes sense that this > patch should modify something there, but: > > * "make check" passes > > * the test warns even before patch 1, for both the DSDT (modified by the > patch) and SSDT (which this series doesn't touch at all) > > * I cannot get it to pass, except by blindly copying the "actual" output > on the "expected" files > > * mst is on vacation and Marcel is off on Fridays > > Based on my understanding of the problem, it is not possible to fix the > bug without hacks like this one, and even reverting all patches in this > area would be more risky. Hmm. I'm not really sure what the right thing is, so what I'm planning to do is: * just apply the qemu-char fix for now * not tag -rc4 today * see if things are clearer on Monday (I see Igor has now sent out a patchset) * tag -rc4 Mon or Tues * slip the release date a few days (not a big deal, I think) thanks -- PMM
Peter Maydell <peter.maydell@linaro.org> wrote: > On 25 July 2014 15:22, Paolo Bonzini <pbonzini@redhat.com> wrote: >> Since Igor hasn't sent his patches, and I'm leaving the office, I pushed >> this to >> >> git://github.com/bonzini/qemu.git tags/for-upstream-full >> >> I don't know about tests/acpi-test-data/pc. It makes sense that this >> patch should modify something there, but: >> >> * "make check" passes >> >> * the test warns even before patch 1, for both the DSDT (modified by the >> patch) and SSDT (which this series doesn't touch at all) >> >> * I cannot get it to pass, except by blindly copying the "actual" output >> on the "expected" files >> >> * mst is on vacation and Marcel is off on Fridays >> >> Based on my understanding of the problem, it is not possible to fix the >> bug without hacks like this one, and even reverting all patches in this >> area would be more risky. > > Hmm. I'm not really sure what the right thing is, so what > I'm planning to do is: > * just apply the qemu-char fix for now > * not tag -rc4 today > * see if things are clearer on Monday (I see Igor has now > sent out a patchset) > * tag -rc4 Mon or Tues > * slip the release date a few days (not a big deal, I think) I am reading both patch-sets. I preffer very much Paolo solution to Igor one. But I have to say that I don't understand PATCH 1 (neither before or after the change). Solution does what we should do, that is generate the size that destination is expecting, and no simply blindy accept packages that are smaller. The compatibility bits of PATCH2 look ok (that ones, I can kind of understand them). Igor? Later, Juan.
On Mon, 28 Jul 2014 14:59:26 +0200 Juan Quintela <quintela@redhat.com> wrote: > Peter Maydell <peter.maydell@linaro.org> wrote: > > On 25 July 2014 15:22, Paolo Bonzini <pbonzini@redhat.com> wrote: > >> Since Igor hasn't sent his patches, and I'm leaving the office, I pushed > >> this to > >> > >> git://github.com/bonzini/qemu.git tags/for-upstream-full > >> > >> I don't know about tests/acpi-test-data/pc. It makes sense that this > >> patch should modify something there, but: > >> > >> * "make check" passes > >> > >> * the test warns even before patch 1, for both the DSDT (modified by the > >> patch) and SSDT (which this series doesn't touch at all) > >> > >> * I cannot get it to pass, except by blindly copying the "actual" output > >> on the "expected" files > >> > >> * mst is on vacation and Marcel is off on Fridays > >> > >> Based on my understanding of the problem, it is not possible to fix the > >> bug without hacks like this one, and even reverting all patches in this > >> area would be more risky. > > > > Hmm. I'm not really sure what the right thing is, so what > > I'm planning to do is: > > * just apply the qemu-char fix for now > > * not tag -rc4 today > > * see if things are clearer on Monday (I see Igor has now > > sent out a patchset) > > * tag -rc4 Mon or Tues > > * slip the release date a few days (not a big deal, I think) > > I am reading both patch-sets. > > I preffer very much Paolo solution to Igor one. > > But I have to say that I don't understand PATCH 1 (neither before or > after the change). Solution does what we should do, that is generate > the size that destination is expecting, and no simply blindy accept > packages that are smaller. > > The compatibility bits of PATCH2 look ok (that ones, I can kind of > understand them). > > > Igor? These patches work for 2.0->2.1 for -M pc-1.7 (Ubuntu case), however they doesn't for 1.7->2.1 (RHEL7 case). Also chasing exact size is a bit problematic considering that different iasl versions produce differently sized AML output. As far as I understood from chat on #qemu, Michael is going to pick-up this series + extendable RAMBlock series + disabling PCI bridge hotplug patch and may be cook additional one to make working backwards migration. > > Later, Juan.
The following changes since commit f368c33d5ab09dd5656924185cd975b11838cd25: Update version for v2.1.0-rc3 release (2014-07-22 18:17:03 +0100) are available in the git repository at: git://github.com/bonzini/qemu.git tags/for-upstream for you to fetch changes up to d975e28437377bc9a65fb5f4f9486a74c7a124cc: qemu-char: ignore flow control if a PTY's slave is not connected (2014-07-24 18:31:50 +0200) ---------------------------------------------------------------- Here are my patches for the last 2.1 rc. ---------------------------------------------------------------- Paolo Bonzini (3): acpi-dsdt: procedurally generate _PRT pc: hack for migration compatibility from QEMU 2.0 qemu-char: ignore flow control if a PTY's slave is not connected hw/i386/acpi-build.c | 75 +- hw/i386/acpi-dsdt.dsl | 90 +- hw/i386/acpi-dsdt.hex.generated | 1910 +++------------------------------------ hw/i386/pc_piix.c | 19 + hw/i386/pc_q35.c | 5 + include/hw/i386/pc.h | 1 + qemu-char.c | 7 + 7 files changed, 250 insertions(+), 1857 deletions(-)