| Message ID | 20220824091837.301708-6-d-tatianin@yandex-team.ru |
|---|---|
| State | New |
| Headers | show |
| Series | vhost-user-blk: dynamically resize config space based on features | expand |
On Wed, Aug 24, 2022 at 12:18:37PM +0300, Daniil Tatianin wrote: > Make vhost-user-blk backwards compatible when migrating from older VMs > running with modern features turned off, the same way it was done for > virtio-blk in 20764be0421c ("virtio-blk: set config size depending on the features enabled") > > It's currently impossible to migrate from an older VM with > vhost-user-blk (with disable-legacy=off) because of errors like this: > > qemu-system-x86_64: get_pci_config_device: Bad config data: i=0x10 read: 41 device: 1 cmask: ff wmask: 80 w1cmask:0 > qemu-system-x86_64: Failed to load PCIDevice:config > qemu-system-x86_64: Failed to load virtio-blk:virtio > qemu-system-x86_64: error while loading state for instance 0x0 of device '0000:00:05.0:00.0:02.0/virtio-blk' > qemu-system-x86_64: load of migration failed: Invalid argument > > This is caused by the newer (destination) VM requiring a bigger BAR0 > alignment because it has to cover a bigger configuration space, which > isn't actually needed since those additional config fields are not > active (write-zeroes/discard). > > Signed-off-by: Daniil Tatianin <d-tatianin@yandex-team.ru> > --- > hw/block/vhost-user-blk.c | 15 ++++++++------- > include/hw/virtio/vhost-user-blk.h | 1 + > 2 files changed, 9 insertions(+), 7 deletions(-) > > diff --git a/hw/block/vhost-user-blk.c b/hw/block/vhost-user-blk.c > index 64f3457373..d18a7a2cd4 100644 > --- a/hw/block/vhost-user-blk.c > +++ b/hw/block/vhost-user-blk.c > @@ -23,6 +23,7 @@ > #include "hw/qdev-core.h" > #include "hw/qdev-properties.h" > #include "hw/qdev-properties-system.h" > +#include "hw/virtio/virtio-blk-common.h" > #include "hw/virtio/vhost.h" > #include "hw/virtio/vhost-user-blk.h" > #include "hw/virtio/virtio.h" > @@ -63,7 +64,7 @@ static void vhost_user_blk_update_config(VirtIODevice *vdev, uint8_t *config) > /* Our num_queues overrides the device backend */ > virtio_stw_p(vdev, &s->blkcfg.num_queues, s->num_queues); > > - memcpy(config, &s->blkcfg, sizeof(struct virtio_blk_config)); > + memcpy(config, &s->blkcfg, s->config_size); Please drop the s->config_size field introduced in this patch and use vdev->config_len instead. When the same value is stored in multiple places it's hard to be sure each copy remains identical and bugs can creep in. For example, if vdev->config_len is used consistently then it's clear that buffer overflows and information leaks are prevented by common code: uint32_t virtio_config_readb(VirtIODevice *vdev, uint32_t addr) { VirtioDeviceClass *k = VIRTIO_DEVICE_GET_CLASS(vdev); uint8_t val; if (addr + sizeof(val) > vdev->config_len) { ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^ return (uint32_t)-1; } k->get_config(vdev, vdev->config); It's safe because vdev->config is g_malloc0(vdev->config_len). Buf if I see s->config_size, I don't really know whether it's safe and I need to audit the code to be sure.
diff --git a/hw/block/vhost-user-blk.c b/hw/block/vhost-user-blk.c index 64f3457373..d18a7a2cd4 100644 --- a/hw/block/vhost-user-blk.c +++ b/hw/block/vhost-user-blk.c @@ -23,6 +23,7 @@ #include "hw/qdev-core.h" #include "hw/qdev-properties.h" #include "hw/qdev-properties-system.h" +#include "hw/virtio/virtio-blk-common.h" #include "hw/virtio/vhost.h" #include "hw/virtio/vhost-user-blk.h" #include "hw/virtio/virtio.h" @@ -63,7 +64,7 @@ static void vhost_user_blk_update_config(VirtIODevice *vdev, uint8_t *config) /* Our num_queues overrides the device backend */ virtio_stw_p(vdev, &s->blkcfg.num_queues, s->num_queues); - memcpy(config, &s->blkcfg, sizeof(struct virtio_blk_config)); + memcpy(config, &s->blkcfg, s->config_size); } static void vhost_user_blk_set_config(VirtIODevice *vdev, const uint8_t *config) @@ -96,8 +97,7 @@ static int vhost_user_blk_handle_config_change(struct vhost_dev *dev) Error *local_err = NULL; ret = vhost_dev_get_config(dev, (uint8_t *)&blkcfg, - sizeof(struct virtio_blk_config), - &local_err); + s->config_size, &local_err); if (ret < 0) { error_report_err(local_err); return ret; @@ -106,7 +106,7 @@ static int vhost_user_blk_handle_config_change(struct vhost_dev *dev) /* valid for resize only */ if (blkcfg.capacity != s->blkcfg.capacity) { s->blkcfg.capacity = blkcfg.capacity; - memcpy(dev->vdev->config, &s->blkcfg, sizeof(struct virtio_blk_config)); + memcpy(dev->vdev->config, &s->blkcfg, s->config_size); virtio_notify_config(dev->vdev); } @@ -444,7 +444,7 @@ static int vhost_user_blk_realize_connect(VHostUserBlk *s, Error **errp) assert(s->connected); ret = vhost_dev_get_config(&s->dev, (uint8_t *)&s->blkcfg, - sizeof(struct virtio_blk_config), errp); + s->config_size, errp); if (ret < 0) { qemu_chr_fe_disconnect(&s->chardev); vhost_dev_cleanup(&s->dev); @@ -489,8 +489,9 @@ static void vhost_user_blk_device_realize(DeviceState *dev, Error **errp) return; } - virtio_init(vdev, VIRTIO_ID_BLOCK, - sizeof(struct virtio_blk_config)); + s->config_size = virtio_blk_common_get_config_size(s->host_features); + + virtio_init(vdev, VIRTIO_ID_BLOCK, s->config_size); s->virtqs = g_new(VirtQueue *, s->num_queues); for (i = 0; i < s->num_queues; i++) { diff --git a/include/hw/virtio/vhost-user-blk.h b/include/hw/virtio/vhost-user-blk.h index 6252095c45..b7810360b9 100644 --- a/include/hw/virtio/vhost-user-blk.h +++ b/include/hw/virtio/vhost-user-blk.h @@ -52,6 +52,7 @@ struct VHostUserBlk { bool started_vu; uint64_t host_features; + size_t config_size; }; #endif
Make vhost-user-blk backwards compatible when migrating from older VMs running with modern features turned off, the same way it was done for virtio-blk in 20764be0421c ("virtio-blk: set config size depending on the features enabled") It's currently impossible to migrate from an older VM with vhost-user-blk (with disable-legacy=off) because of errors like this: qemu-system-x86_64: get_pci_config_device: Bad config data: i=0x10 read: 41 device: 1 cmask: ff wmask: 80 w1cmask:0 qemu-system-x86_64: Failed to load PCIDevice:config qemu-system-x86_64: Failed to load virtio-blk:virtio qemu-system-x86_64: error while loading state for instance 0x0 of device '0000:00:05.0:00.0:02.0/virtio-blk' qemu-system-x86_64: load of migration failed: Invalid argument This is caused by the newer (destination) VM requiring a bigger BAR0 alignment because it has to cover a bigger configuration space, which isn't actually needed since those additional config fields are not active (write-zeroes/discard). Signed-off-by: Daniil Tatianin <d-tatianin@yandex-team.ru> --- hw/block/vhost-user-blk.c | 15 ++++++++------- include/hw/virtio/vhost-user-blk.h | 1 + 2 files changed, 9 insertions(+), 7 deletions(-)