similar to: [PATCH 1/2] virtio_pci: double free and invalid memory access of device vqs

This is based on Sasha's patch, with some tweaks. Michael S. Tsirkin (2): virtio_pci: device-specific release callback virtio_pci: document why we defer kfree Sasha Levin (1): virtio_pci: defer kfree until release callback drivers/virtio/virtio_pci_common.h | 1 - drivers/virtio/virtio_pci_common.c | 9 --------- drivers/virtio/virtio_pci_legacy.c | 12 +++++++++++- 3 files

This is based on Sasha's patch, with some tweaks. Michael S. Tsirkin (2): virtio_pci: device-specific release callback virtio_pci: document why we defer kfree Sasha Levin (1): virtio_pci: defer kfree until release callback drivers/virtio/virtio_pci_common.h | 1 - drivers/virtio/virtio_pci_common.c | 9 --------- drivers/virtio/virtio_pci_legacy.c | 12 +++++++++++- 3 files

Move resource allocation from common code to legacy and modern code. Only request resources actually used, i.e. bar0 in legacy mode and the bar(s) specified by capabilities in modern mode. Signed-off-by: Gerd Hoffmann <kraxel at redhat.com> --- drivers/virtio/virtio_pci_common.c | 7 ------- drivers/virtio/virtio_pci_common.h | 2 ++ drivers/virtio/virtio_pci_legacy.c | 13 ++++++++++++-

Move resource allocation from common code to legacy and modern code. Only request resources actually used, i.e. bar0 in legacy mode and the bar(s) specified by capabilities in modern mode. Signed-off-by: Gerd Hoffmann <kraxel at redhat.com> --- drivers/virtio/virtio_pci_common.c | 7 ------- drivers/virtio/virtio_pci_common.h | 2 ++ drivers/virtio/virtio_pci_legacy.c | 13 ++++++++++++-

Move resource allocation from common code to legacy and modern code. Only request resources actually used, i.e. bar0 in legacy mode and the bar(s) specified by capabilities in modern mode. Signed-off-by: Gerd Hoffmann <kraxel at redhat.com> --- drivers/virtio/virtio_pci_common.c | 7 ------- drivers/virtio/virtio_pci_common.h | 2 ++ drivers/virtio/virtio_pci_legacy.c | 13 ++++++++++++-

Move resource allocation from common code to legacy and modern code. Only request resources actually used, i.e. bar0 in legacy mode and the bar(s) specified by capabilities in modern mode. Signed-off-by: Gerd Hoffmann <kraxel at redhat.com> --- drivers/virtio/virtio_pci_common.c | 7 ------- drivers/virtio/virtio_pci_common.h | 2 ++ drivers/virtio/virtio_pci_legacy.c | 13 ++++++++++++-

This is just repost of all patches with fixups squashed in - convenient if you just want to remove old ones from queue and apply new ones. I also tweaked commit log for patch "virtio_pci: modern driver" I also included Gerd's tag: Tested-by: Gerd Hoffmann <kraxel at redhat.com> You can find it all before the rebase -i --autosquash in my tree:

Move resource allocation from common code to legacy and modern code. Only request resources actually used, i.e. bar0 in legacy mode and the regions specified by capabilities in modern mode. Signed-off-by: Gerd Hoffmann <kraxel at redhat.com> --- drivers/virtio/virtio_pci_common.c | 7 ----- drivers/virtio/virtio_pci_common.h | 5 ++++ drivers/virtio/virtio_pci_legacy.c | 13 +++++++++-

This is just repost of all patches with fixups squashed in - convenient if you just want to remove old ones from queue and apply new ones. I also tweaked commit log for patch "virtio_pci: modern driver" I also included Gerd's tag: Tested-by: Gerd Hoffmann <kraxel at redhat.com> You can find it all before the rebase -i --autosquash in my tree:

Move resource allocation from common code to legacy and modern code. Only request resources actually used, i.e. bar0 in legacy mode and the regions specified by capabilities in modern mode. Signed-off-by: Gerd Hoffmann <kraxel at redhat.com> --- drivers/virtio/virtio_pci_common.c | 7 ----- drivers/virtio/virtio_pci_common.h | 5 ++++ drivers/virtio/virtio_pci_legacy.c | 13 +++++++++-

Move resource allocation from common code to legacy and modern code. Only request resources actually used, i.e. bar0 in legacy mode and the regions specified by capabilities in modern mode. Signed-off-by: Gerd Hoffmann <kraxel at redhat.com> --- drivers/virtio/virtio_pci_common.c | 7 ---- drivers/virtio/virtio_pci_common.h | 5 +++ drivers/virtio/virtio_pci_legacy.c | 13 +++++++-

Move resource allocation from common code to legacy and modern code. Only request resources actually used, i.e. bar0 in legacy mode and the regions specified by capabilities in modern mode. Signed-off-by: Gerd Hoffmann <kraxel at redhat.com> --- drivers/virtio/virtio_pci_common.c | 7 ---- drivers/virtio/virtio_pci_common.h | 5 +++ drivers/virtio/virtio_pci_legacy.c | 13 +++++++-

This is just the full patchset reposted with fixups in correct order, before squashing them. I also tweaked commit log for patch "virtio_pci: modern driver" I also included Gerd's tag: Tested-by: Gerd Hoffmann <kraxel at redhat.com> You can find it all before the rebase -i --autosquash in my tree: git://git.kernel.org/pub/scm/linux/kernel/git/mst/vhost.git virtio-next

This is just the full patchset reposted with fixups in correct order, before squashing them. I also tweaked commit log for patch "virtio_pci: modern driver" I also included Gerd's tag: Tested-by: Gerd Hoffmann <kraxel at redhat.com> You can find it all before the rebase -i --autosquash in my tree: git://git.kernel.org/pub/scm/linux/kernel/git/mst/vhost.git virtio-next

On Fri, Jan 02, 2015 at 02:47:40PM -0500, Sasha Levin wrote: > When releasing a virtio device, We can't free a struct virtio_device until the > underlying struct device has been released, which might not happen immediately > on device_unregister() even if that was the device's last reference. > > Instead, free the memory only once we know the device is gone in the release

On Fri, Jan 02, 2015 at 02:47:40PM -0500, Sasha Levin wrote: > When releasing a virtio device, We can't free a struct virtio_device until the > underlying struct device has been released, which might not happen immediately > on device_unregister() even if that was the device's last reference. > > Instead, free the memory only once we know the device is gone in the release

Changes since v2: handling for devices without config space (e.g. rng) reduce # of mappings for VQs These patches seem to work fine on my virtio-1.0 qemu branch. There haven't been any bugs since v2: just minor cleanups and enhancements. QEMU side is still undergoing polishing, but is already testable. Rusty, what do you think? Let's merge these for 3.20? Also - will you be doing that

Changes since v2: handling for devices without config space (e.g. rng) reduce # of mappings for VQs These patches seem to work fine on my virtio-1.0 qemu branch. There haven't been any bugs since v2: just minor cleanups and enhancements. QEMU side is still undergoing polishing, but is already testable. Rusty, what do you think? Let's merge these for 3.20? Also - will you be doing that

Fixes a couple of minor compliance issues in new virtio 1.0 code. Plus, adds a couple of minor cleanups - not bugfixes, but seem safe enough for 3.19. Michael S. Tsirkin (6): virtio: set VIRTIO_CONFIG_S_FEATURES_OK on restore virtio_config: fix virtio_cread_bytes virtio_pci_common.h: drop VIRTIO_PCI_NO_LEGACY virtio_pci: move probe to common file virtio_pci: add VIRTIO_PCI_NO_LEGACY

Fixes a couple of minor compliance issues in new virtio 1.0 code. Plus, adds a couple of minor cleanups - not bugfixes, but seem safe enough for 3.19. Michael S. Tsirkin (6): virtio: set VIRTIO_CONFIG_S_FEATURES_OK on restore virtio_config: fix virtio_cread_bytes virtio_pci_common.h: drop VIRTIO_PCI_NO_LEGACY virtio_pci: move probe to common file virtio_pci: add VIRTIO_PCI_NO_LEGACY