Planning/4.1: Difference between revisions

Latest revision as of 20:51, 5 November 2019

Release Schedule

2018-04-24	Beginning of development phase
2019-07-02	Soft feature freeze. Only bug fixes after this point. All feature changes must be already in a sub maintainer tree and all pull requests from submaintainers must have been sent to the list by this date.
2019-07-09	Hard feature freeze. Tag rc0
2019-07-16	Tag rc1
2019-07-23	Tag rc2
2019-07-30	Tag rc3
2019-08-06	Tag rc4
2019-08-13	Tag rc5
2019-08-15	Release
2019-11-12	Freeze for 4.1.1
2019-11-14	Tag v4.1.1

Known issues

Please list all patch series or bugs which need to be fixed for the release here. That way I can be sure I don't miss any when I'm rolling release candidates.

Fixed in rc0

Fixed in rc1

virtio-balloon/migration compatibility: There's a migration break between 3.1<->4.x; reported by Wolfgang Bumiller; fixed by stefan's virtio-balloon: fix QEMU 4.0 config size migration incompatibility 2019-07-10 - fixed by 2bbadb08ce272d65e1f78621002008b07d1e0f03
migration: migratie/migrate_cancel/migrate: Ivan Ren's post from 14th July - 'migration: always initial RAMBlock.bmap to 1 for new migration' - fixed by 40c4d4a835453452a262f32450a0449886aa19ce

Fixed in rc2

x86 --without-default-devices is broken due to CONFIG_VMMOUSE being "select"ed.
The virtio-pmem device is no longer "at-risk" -- the kernel header changes got taken into mainline without changes that require us to resync or to revert the QEMU side changes

Fixed in rc3

snowridge cpu name/features
coverity issue in block/nvme
virtio devices seem to have lost PCIe capability and PCIe powermanagement capabilities (David Gilbert posted revert's for 'report invalid disable-legacy|modern usage for virtio-1-only devs')
copy-on-read migration permissions issue
inserting read-only image to initially empty scsi-cd is incorrectly rejected
inserting read-only image to initially empty floppy drive is incorrectly rejected
arm emulated BKPT/BRK insn at EL2 can be taken to EL1
unwanted ACS capability on PCI bridges on old machine types

Fixed in rc4

M-profile CPUs with FPU but without Security Extensions do bogus NSACR checks (LP:1838475)
We forgot to update the slirp submodule
The backup job’s use of copy offloading is broken
A mirror job with copy-mode=write-blocking may fail to copy some data from the source
Performing an incremental backup with a bitmap that had been added to a block node before it was grown with block_resize will ignore all dirty areas past the node’s original size
Compat handling turns off 'edid' only for virtio-gpu-pci, but not for virtio-gpu-ccw

Fixed in rc5

PPC bugs with the XIVE interrupt controller for spapr
Makefile issue if you're building the risc-v bios blobs from source
Security bug with bochs-display device

Not yet fixed in any rc

Issues that will not be fixed

s390x TCG: The VERIM instruction can make QEMU crash

Targeted Features

See the ChangeLog/4.1 for full details.

@@ Line 1: / Line 1: @@
 == Release Schedule ==
-Tentatively proposed dates, still to be discussed/agreed.
 {| border="1"
@@ Line 23: / Line 21: @@
 |-
 | 2019-08-06
-| Release, or tag rc4 if extra RC cycle needed
+| Tag rc4
 |-
 | 2019-08-13
-| Release, if we needed an rc4.
+| Tag rc5
+|-
+| 2019-08-15
+| Release
+|-
+| 2019-11-12
+| Freeze for 4.1.1
+|-
+| 2019-11-14
+| Tag v4.1.1
 |}
@@ Line 34: / Line 41: @@
 === Fixed in rc0 ===
+=== Fixed in rc1 ===
+* virtio-balloon/migration compatibility: There's a migration break between 3.1<->4.x; reported by Wolfgang Bumiller; fixed by stefan's virtio-balloon: fix QEMU 4.0 config size migration incompatibility  2019-07-10 - fixed by 2bbadb08ce272d65e1f78621002008b07d1e0f03
+* migration: migratie/migrate_cancel/migrate: Ivan Ren's post from 14th July - 'migration: always initial RAMBlock.bmap to 1 for new migration' - fixed by 40c4d4a835453452a262f32450a0449886aa19ce
+=== Fixed in rc2 ===
+* x86 --without-default-devices is broken due to CONFIG_VMMOUSE being "select"ed.
+* The virtio-pmem device is no longer "at-risk" -- the kernel header changes got taken into mainline without changes that require us to resync or to revert the QEMU side changes
+=== Fixed in rc3 ===
+* snowridge cpu name/features
+* coverity issue in block/nvme
+* virtio devices seem to have lost PCIe capability and PCIe powermanagement capabilities (David Gilbert posted revert's for 'report invalid disable-legacy|modern usage for virtio-1-only devs')
+* copy-on-read migration permissions issue
+* inserting read-only image to initially empty scsi-cd is incorrectly rejected
+* inserting read-only image to initially empty floppy drive is incorrectly rejected
+* arm emulated BKPT/BRK insn at EL2 can be taken to EL1
+* unwanted ACS capability on PCI bridges on old machine types
+=== Fixed in rc4 ===
+* M-profile CPUs with FPU but without Security Extensions do bogus NSACR checks (LP:1838475)
+* We forgot to update the slirp submodule
+* The backup job’s use of copy offloading is broken
+* A mirror job with copy-mode=write-blocking may fail to copy some data from the source
+* Performing an incremental backup with a bitmap that had been added to a block node before it was grown with block_resize will ignore all dirty areas past the node’s original size
+* Compat handling turns off 'edid' only for virtio-gpu-pci, but not for virtio-gpu-ccw
+=== Fixed in rc5 ===
+* PPC bugs with the XIVE interrupt controller for spapr
+* Makefile issue if you're building the risc-v bios blobs from source
+* Security bug with bochs-display device
 === Not yet fixed in any rc ===
-* virtio-balloon/migration compatibility: There's a migration break between 3.1<->4.x; reported by Wolfgang Bumiller; fixed by stefan's virtio-balloon: fix QEMU 4.0 config size migration incompatibility  2019-07-10
-* migration: migratie/migrate_cancel/migrate: Ivan Ren's post from 14th July - 'migration: always initial RAMBlock.bmap to 1 for new migration'
 === Issues that will not be fixed ===
+* s390x TCG: The VERIM instruction can make QEMU crash
 == Targeted Features ==
 See the [[ChangeLog/4.1]] for full details.