Fix overflow bug in qcow2 discard
v5 cover Fix overflow bug in qcow2 discard
v5: by Kevin's comments
02: add check for invalid @bytes parameter, drop r-b
03: move from qemu-img info to qemu-img map -f raw, drop r-b and t-b
Vladimir Sementsov-Ogievskiy (3):
block/qcow2-refcount: add trace-point to qcow2_process_discards block/io: bdrv_pdiscard: support int64_t bytes parameter
iotests: test big qcow2 shrink
include/block/block.h | 4 +-- block/io.c | 16 ++++----- block/qcow2-refcount.c | 7 +++- block/trace-events | 3 ++ tests/qemu-iotests/250 | 73 ++++++++++++++++++++++++++++++++++++++ tests/qemu-iotests/250.out | 23 ++++++++++++ tests/qemu-iotests/group | 1 + 7 files changed, 116 insertions(+), 11 deletions(-) create mode 100755 tests/qemu-iotests/250 create mode 100644 tests/qemu-iotests/250.out
Signature · 1 line
2.18.0
PATCH v5 1/3 block/qcow2-refcount: add trace-point to qcow2_process_discards
Let's at least trace ignored failure.
--- block/qcow2-refcount.c | 7 ++++++- block/trace-events | 3 +++ 2 files changed, 9 insertions(+), 1 deletion(-)
block/qcow2-refcount.c +6 -1
@@ -30,6 +30,7 @@#include "qemu/range.h" #include "qemu/bswap.h" #include "qemu/cutils.h" +#include "trace.h" static int64_t alloc_clusters_noref (BlockDriverState *bs,uint64_t size,uint64_t max);
@@ -738,7 +739,11 @@ void qcow2_process_discards(BlockDriverState *bs, int ret)/* Discard is optional, ignore the return value */ if (ret >=0 ) { -bdrv_pdiscard (bs->file, d->offset, d->bytes); +int r2 =bdrv_pdiscard (bs->file, d->offset, d->bytes); +if (r2 <0 ) { +trace_qcow2_process_discards_failed_region (d->offset, d->bytes, + r2); + } }g_free (d);
block/trace-events +3 -0
@@ -91,6 +91,9 @@ qcow2_cache_get_done(void *co, int c, int i) "co %p is_l2_cache %d index %d" qcow2_cache_flush(void *co, int c) "co %p is_l2_cache %d" qcow2_cache_entry_flush(void *co, int c, int i) "co %p is_l2_cache %d index %d" +# qcow2-refcount.c +qcow2_process_discards_failed_region(uint64_t offset, uint64_t bytes, int ret) "offset 0x%" PRIx64 " bytes 0x%" PRIx64 " ret %d" + # qed-l2-cache.c qed_alloc_l2_cache_entry(void *l2_cache, void *entry) "l2_cache %p entry %p" qed_unref_l2_cache_entry(void *entry, int ref) "entry %p ref %d"
Signature · 1 line
2.18.0
PATCH v5 2/3 block/io: bdrv_pdiscard: support int64_t bytes parameter
In reply to Vladimir Sementsov-OgievskiyThis fixes at least one overflow in qcow2_process_discards, which passes 64bit region length to bdrv_pdiscard where bytes (or sectors in the past) parameter is int since its introduction in 0b919fae.
--- include/block/block.h | 4 ++-- block/io.c | 16 ++++++++-------- 2 files changed, 10 insertions(+), 10 deletions(-)
include/block/block.h +2 -2
@@ -432,8 +432,8 @@ void bdrv_drain_all(void);AIO_WAIT_WHILE (bdrv_get_aio_context (bs_), \ cond); }) -int bdrv_pdiscard (BdrvChild *child,int64_t offset,int bytes); -int bdrv_co_pdiscard (BdrvChild *child,int64_t offset,int bytes); +int bdrv_pdiscard (BdrvChild *child,int64_t offset,int64_t bytes); +int bdrv_co_pdiscard (BdrvChild *child,int64_t offset,int64_t bytes);int bdrv_has_zero_init_1 (BlockDriverState *bs);int bdrv_has_zero_init (BlockDriverState *bs);bool bdrv_unallocated_blocks_are_zero (BlockDriverState *bs);
block/io.c +8 -8
@@ -2653,7 +2653,7 @@ int bdrv_flush(BlockDriverState *bs)typedef struct DiscardCo {BdrvChild *child;int64_t offset; -int bytes; +int64_t bytes;int ret; }DiscardCo ; static void coroutine_fn bdrv_pdiscard_co_entry(void *opaque)
@@ -2664,14 +2664,15 @@ static void coroutine_fn bdrv_pdiscard_co_entry(void *opaque)aio_wait_kick (); } -int coroutine_fn bdrv_co_pdiscard(BdrvChild *child, int64_t offset, int bytes) +int coroutine_fn bdrv_co_pdiscard(BdrvChild *child, int64_t offset, +int64_t bytes) {BdrvTrackedRequest req;int max_pdiscard, ret;int head, tail, align;BlockDriverState *bs = child->bs; -if (!bs || !bs->drv) { +if (!bs || !bs->drv || !bdrv_is_inserted (bs)) {return -ENOMEDIUM ; }
@@ -2679,9 +2680,8 @@ int coroutine_fn bdrv_co_pdiscard(BdrvChild *child, int64_t offset, int bytes)return -EPERM ; } - ret =bdrv_check_byte_request (bs, offset, bytes); -if (ret <0 ) { -return ret; +if (offset <0 || bytes <0 || bytes >INT64_MAX - offset) { +return -EIO ; }/* Do nothing if disabled. */
@@ -2716,7 +2716,7 @@ int coroutine_fn bdrv_co_pdiscard(BdrvChild *child, int64_t offset, int bytes)assert (max_pdiscard >= bs->bl.request_alignment);while (bytes >0 ) { -int num = bytes; +int64_t num = bytes;if (head) {/* Make small requests to get to alignment boundaries. */
@@ -2778,7 +2778,7 @@ out:return ret; } -int bdrv_pdiscard (BdrvChild *child,int64_t offset,int bytes) +int bdrv_pdiscard (BdrvChild *child,int64_t offset,int64_t bytes) {Coroutine *co;DiscardCo rwco = {
Signature · 1 line
2.18.0
Re: [Qemu-devel] [PATCH v5 2/3] block/io: bdrv_pdiscard: support int64_t bytes parameter
Quoted 43 lines · Vladimir Sementsov-Ogievskiy
This fixes at least one overflow in
qcow2_process_discards, which passes 64bit region length tobdrv_pdiscardwhere bytes (or sectors in the past) parameter is int since its introduction in0b919fae.Signed-off-by Vladimir Sementsov-Ogievskiy <[email protected]>--- include/block/block.h | 4 ++-- block/io.c | 16 ++++++++-------- 2 files changed, 10 insertions(+), 10 deletions(-)include/block/block.h +2
-2@@ -432,8 +432,8 @@ void bdrv_drain_all(void);AIO_WAIT_WHILE (bdrv_get_aio_context (bs_), \ cond); }) -int bdrv_pdiscard (BdrvChild *child,int64_t offset,int bytes); -int bdrv_co_pdiscard (BdrvChild *child,int64_t offset,int bytes); +int bdrv_pdiscard (BdrvChild *child,int64_t offset,int64_t bytes); +int bdrv_co_pdiscard (BdrvChild *child,int64_t offset,int64_t bytes);int bdrv_has_zero_init_1 (BlockDriverState *bs);int bdrv_has_zero_init (BlockDriverState *bs);bool bdrv_unallocated_blocks_are_zero (BlockDriverState *bs);block/io.c +1
-2@@ -2653,7 +2653,7 @@ int bdrv_flush(BlockDriverState *bs)typedef struct DiscardCo {BdrvChild *child;int64_t offset; -int bytes; +int64_t bytes;int ret; }DiscardCo ; static void coroutine_fn bdrv_pdiscard_co_entry(void *opaque)@@ -2664,14 +2664,15 @@ static void coroutine_fn bdrv_pdiscard_co_entry(void *opaque)aio_wait_kick (); } -int coroutine_fn bdrv_co_pdiscard(BdrvChild *child, int64_t offset, int bytes)
block/io.c static void coroutine_fn bdrv_pdiscard_co_entry(void *opaque) · line 2675
Quoted 10 lines
block/io.c +3
-1+int coroutine_fn bdrv_co_pdiscard(BdrvChild *child, int64_t offset, +int64_t bytes) {BdrvTrackedRequest req;int max_pdiscard, ret;int head, tail, align;BlockDriverState *bs = child->bs; -if (!bs || !bs->drv) { +if (!bs || !bs->drv || !bdrv_is_inserted (bs)) {
Should we describe this change in the commit message?
IIUC you added this check because you removed bdrv_check_byte_request() below,
Maybe we can also remove '!bs->drv', since it is checked in bdrv_is_inserted().
block/io.c int coroutine_fn bdrv_co_pdiscard(BdrvChild *child, int64_t offset, int bytes) · line 2685
Quoted 13 lines
block/io.c +2
-3return -ENOMEDIUM ; }@@ -2679,9 +2680,8 @@ int coroutine_fn bdrv_co_pdiscard(BdrvChild *child, int64_t offset, int bytes)return -EPERM ; } - ret =bdrv_check_byte_request (bs, offset, bytes); -if (ret <0 ) { -return ret; +if (offset <0 || bytes <0 || bytes >INT64_MAX - offset) { +return -EIO ; }
Should we check if 'bytes' is greater than
'BDRV_REQUEST_MAX_SECTORS << BDRV_SECTOR_BITS'?
Thanks,
Stefano
Re: [Qemu-devel] [PATCH v5 2/3] block/io: bdrv_pdiscard: support int64_t bytes parameter
Quoted 52 lines · Stefano Garzarella
Quoted 51 lines · Vladimir Sementsov-Ogievskiy
This fixes at least one overflow in
qcow2_process_discards, which passes 64bit region length tobdrv_pdiscardwhere bytes (or sectors in the past) parameter is int since its introduction in0b919fae.Signed-off-by Vladimir Sementsov-Ogievskiy <[email protected]>--- include/block/block.h | 4 ++-- block/io.c | 16 ++++++++-------- 2 files changed, 10 insertions(+), 10 deletions(-)include/block/block.h +2
-2@@ -432,8 +432,8 @@ void bdrv_drain_all(void);AIO_WAIT_WHILE (bdrv_get_aio_context (bs_), \ cond); }) -int bdrv_pdiscard (BdrvChild *child,int64_t offset,int bytes); -int bdrv_co_pdiscard (BdrvChild *child,int64_t offset,int bytes); +int bdrv_pdiscard (BdrvChild *child,int64_t offset,int64_t bytes); +int bdrv_co_pdiscard (BdrvChild *child,int64_t offset,int64_t bytes);int bdrv_has_zero_init_1 (BlockDriverState *bs);int bdrv_has_zero_init (BlockDriverState *bs);bool bdrv_unallocated_blocks_are_zero (BlockDriverState *bs);block/io.c +3
-2@@ -2653,7 +2653,7 @@ int bdrv_flush(BlockDriverState *bs)typedef struct DiscardCo {BdrvChild *child;int64_t offset; -int bytes; +int64_t bytes;int ret; }DiscardCo ; static void coroutine_fn bdrv_pdiscard_co_entry(void *opaque)@@ -2664,14 +2664,15 @@ static void coroutine_fn bdrv_pdiscard_co_entry(void *opaque)aio_wait_kick (); } -int coroutine_fn bdrv_co_pdiscard(BdrvChild *child, int64_t offset, int bytes) +int coroutine_fn bdrv_co_pdiscard(BdrvChild *child, int64_t offset, +int64_t bytes) {BdrvTrackedRequest req;int max_pdiscard, ret;int head, tail, align;BlockDriverState *bs = child->bs;
Quoted 4 lines
block/io.c static void coroutine_fn bdrv_pdiscard_co_entry(void *opaque) · line 2675
Quoted 2 lines
block/io.c +1
-1-if (!bs || !bs->drv) { +if (!bs || !bs->drv || !bdrv_is_inserted (bs)) {Should we describe this change in the commit message?
Honestly, don't want to resend the series for this.
Quoted 5 lines
IIUC you added this check because you removed
bdrv_check_byte_request()below,Maybe we can also remove '!bs->drv', since it is checked in
bdrv_is_inserted().
Hmm, on v4 Kevin commented, that bdrv_is_inserted not needed, and, as I understand, not only in bdrv_co_pdiscard it should be removed, but it may be done later. So, I'd prefer to keep it as is for now.
Quoted 8 lines
Quoted 8 lines
block/io.c +0
-0return -ENOMEDIUM ; }@@ -2679,9 +2680,8 @@ int coroutine_fn bdrv_co_pdiscard(BdrvChild *child, int64_t offset, int bytes)return -EPERM ; }
Quoted 10 lines
block/io.c int coroutine_fn bdrv_co_pdiscard(BdrvChild *child, int64_t offset, int bytes) · line 2685
Quoted 6 lines
block/io.c +2
-3- ret =bdrv_check_byte_request (bs, offset, bytes); -if (ret <0 ) { -return ret; +if (offset <0 || bytes <0 || bytes >INT64_MAX - offset) { +return -EIO ; }Should we check if 'bytes' is greater than
'BDRV_REQUEST_MAX_SECTORS<<BDRV_SECTOR_BITS'?
No, as we are contrariwise trying to support large bytes parameter in bdrv_co_pdiscard, which will exceed max request. If @bytes is large, it will be divided into several smaller requests in the following loop.
Signature · 2 lines
Best regards, Vladimir
Re: [Qemu-devel] [PATCH v5 2/3] block/io: bdrv_pdiscard: support int64_t bytes parameter
Am 30.04.2019 um 12:03 hat Vladimir Sementsov-Ogievskiy geschrieben:
Quoted 53 lines
Quoted 52 lines · Stefano Garzarella
Quoted 51 lines · Vladimir Sementsov-Ogievskiy
This fixes at least one overflow in
qcow2_process_discards, which passes 64bit region length tobdrv_pdiscardwhere bytes (or sectors in the past) parameter is int since its introduction in0b919fae.Signed-off-by Vladimir Sementsov-Ogievskiy <[email protected]>--- include/block/block.h | 4 ++-- block/io.c | 16 ++++++++-------- 2 files changed, 10 insertions(+), 10 deletions(-)include/block/block.h +2
-2@@ -432,8 +432,8 @@ void bdrv_drain_all(void);AIO_WAIT_WHILE (bdrv_get_aio_context (bs_), \ cond); }) -int bdrv_pdiscard (BdrvChild *child,int64_t offset,int bytes); -int bdrv_co_pdiscard (BdrvChild *child,int64_t offset,int bytes); +int bdrv_pdiscard (BdrvChild *child,int64_t offset,int64_t bytes); +int bdrv_co_pdiscard (BdrvChild *child,int64_t offset,int64_t bytes);int bdrv_has_zero_init_1 (BlockDriverState *bs);int bdrv_has_zero_init (BlockDriverState *bs);bool bdrv_unallocated_blocks_are_zero (BlockDriverState *bs);block/io.c +3
-2@@ -2653,7 +2653,7 @@ int bdrv_flush(BlockDriverState *bs)typedef struct DiscardCo {BdrvChild *child;int64_t offset; -int bytes; +int64_t bytes;int ret; }DiscardCo ; static void coroutine_fn bdrv_pdiscard_co_entry(void *opaque)@@ -2664,14 +2664,15 @@ static void coroutine_fn bdrv_pdiscard_co_entry(void *opaque)aio_wait_kick (); } -int coroutine_fn bdrv_co_pdiscard(BdrvChild *child, int64_t offset, int bytes) +int coroutine_fn bdrv_co_pdiscard(BdrvChild *child, int64_t offset, +int64_t bytes) {BdrvTrackedRequest req;int max_pdiscard, ret;int head, tail, align;BlockDriverState *bs = child->bs;
Quoted 6 lines
Quoted 4 lines
block/io.c static void coroutine_fn bdrv_pdiscard_co_entry(void *opaque) · line 2675
Quoted 2 lines
block/io.c +1
-1-if (!bs || !bs->drv) { +if (!bs || !bs->drv || !bdrv_is_inserted (bs)) {Should we describe this change in the commit message?
Honestly, don't want to resend the series for this.
I haven't reviewed the patches yet, but if this remains the only thing to change, it can be updated while applying the series if we have a specific proposal for a new commit message.
Kevin
Eric Blake
Reviewed-byRe: [Qemu-devel] [PATCH v5 2/3] block/io: bdrv_pdiscard: support int64_t bytes parameter
Quoted 12 lines · Kevin Wolf
Am 30.04.2019 um 12:03 hat Vladimir Sementsov-Ogievskiy geschrieben:
Quoted 11 lines
Quoted 10 lines · Stefano Garzarella
Quoted 8 lines · Vladimir Sementsov-Ogievskiy
This fixes at least one overflow in
qcow2_process_discards, which passes 64bit region length tobdrv_pdiscardwhere bytes (or sectors in the past) parameter is int since its introduction in0b919fae.Signed-off-by Vladimir Sementsov-Ogievskiy <[email protected]>block/io.c +1
-1-if (!bs || !bs->drv) { +if (!bs || !bs->drv || !bdrv_is_inserted (bs)) {
Quoted 7 lines
Quoted 3 lines
Quoted 1 line
Should we describe this change in the commit message?
Honestly, don't want to resend the series for this.
I haven't reviewed the patches yet, but if this remains the only thing to change, it can be updated while applying the series if we have a specific proposal for a new commit message.
How about:
This fixes at least one overflow in qcow2_process_discards, which was inadvertently truncating a 64-bit region length to the bdrv_pdiscard 'int bytes' parameter (previously bdrv_discard's 'int sectors') since its introduction in 0b919fae.
By inlining the remaining portions of bdrv_check_byte_request() that are still independent from the code previously validating a 32-bit request, we can drop the call to that function. A request larger than 31 bits (or the driver's max discard limit, if specified) is still split into smaller chunks by the block layer before reaching the driver.
At any rate, you can add:
Signature · 3 lines
Eric Blake, Principal Software Engineer Red Hat, Inc. +1-919-301-3226 Virtualization: qemu.org | libvirt.org
Re: [Qemu-devel] [PATCH v5 2/3] block/io: bdrv_pdiscard: support int64_t bytes parameter
Quoted 28 lines · Eric Blake
Quoted 12 lines · Kevin Wolf
Am 30.04.2019 um 12:03 hat Vladimir Sementsov-Ogievskiy geschrieben:
Quoted 11 lines
Quoted 10 lines · Stefano Garzarella
Quoted 8 lines · Vladimir Sementsov-Ogievskiy
This fixes at least one overflow in
qcow2_process_discards, which passes 64bit region length tobdrv_pdiscardwhere bytes (or sectors in the past) parameter is int since its introduction in0b919fae.Signed-off-by Vladimir Sementsov-Ogievskiy <[email protected]>block/io.c +1
-1-if (!bs || !bs->drv) { +if (!bs || !bs->drv || !bdrv_is_inserted (bs)) {Quoted 7 lines
Quoted 3 lines
Quoted 1 line
Should we describe this change in the commit message?
Honestly, don't want to resend the series for this.
I haven't reviewed the patches yet, but if this remains the only thing to change, it can be updated while applying the series if we have a specific proposal for a new commit message.
How about:
This fixes at least one overflow in
qcow2_process_discards, which was inadvertently truncating a 64-bit region length to thebdrv_pdiscard'int bytes' parameter (previously bdrv_discard's 'int sectors') since its introduction in0b919fae.
Quoted 6 lines
By inlining the remaining portions of
bdrv_check_byte_request()that are still independent from the code previously validating a 32-bit request, we can drop the call to that function. A request larger than 31 bits (or the driver's max discard limit, if specified) is still split into smaller chunks by the block layer before reaching the driver.
That sounds good to me!
Thanks,
Stefano
Re: [Qemu-devel] [PATCH v5 2/3] block/io: bdrv_pdiscard: support int64_t bytes parameter
Quoted 29 lines · Stefano Garzarella
Quoted 28 lines · Eric Blake
Quoted 12 lines · Kevin Wolf
Am 30.04.2019 um 12:03 hat Vladimir Sementsov-Ogievskiy geschrieben:
Quoted 11 lines
Quoted 10 lines · Stefano Garzarella
Quoted 8 lines · Vladimir Sementsov-Ogievskiy
This fixes at least one overflow in
qcow2_process_discards, which passes 64bit region length tobdrv_pdiscardwhere bytes (or sectors in the past) parameter is int since its introduction in0b919fae.Signed-off-by Vladimir Sementsov-Ogievskiy <[email protected]>block/io.c +1
-1-if (!bs || !bs->drv) { +if (!bs || !bs->drv || !bdrv_is_inserted (bs)) {Quoted 7 lines
Quoted 3 lines
Quoted 1 line
Should we describe this change in the commit message?
Honestly, don't want to resend the series for this.
I haven't reviewed the patches yet, but if this remains the only thing to change, it can be updated while applying the series if we have a specific proposal for a new commit message.
How about:
This fixes at least one overflow in
qcow2_process_discards, which was inadvertently truncating a 64-bit region length to thebdrv_pdiscard'int bytes' parameter (previously bdrv_discard's 'int sectors') since its introduction in0b919fae.
Quoted 9 lines
Quoted 6 lines
By inlining the remaining portions of
bdrv_check_byte_request()that are still independent from the code previously validating a 32-bit request, we can drop the call to that function. A request larger than 31 bits (or the driver's max discard limit, if specified) is still split into smaller chunks by the block layer before reaching the driver.That sounds good to me!
And for me. Thank you, Eric!
Signature · 2 lines
Best regards, Vladimir
Stefano Garzarella
Reviewed-byRe: [Qemu-devel] [PATCH v5 2/3] block/io: bdrv_pdiscard: support int64_t bytes parameter
In reply to Vladimir Sementsov-OgievskiyQuoted 63 lines · Vladimir Sementsov-Ogievskiy
Quoted 52 lines · Stefano Garzarella
Quoted 51 lines · Vladimir Sementsov-Ogievskiy
This fixes at least one overflow in
qcow2_process_discards, which passes 64bit region length tobdrv_pdiscardwhere bytes (or sectors in the past) parameter is int since its introduction in0b919fae.Signed-off-by Vladimir Sementsov-Ogievskiy <[email protected]>--- include/block/block.h | 4 ++-- block/io.c | 16 ++++++++-------- 2 files changed, 10 insertions(+), 10 deletions(-)include/block/block.h +2
-2@@ -432,8 +432,8 @@ void bdrv_drain_all(void);AIO_WAIT_WHILE (bdrv_get_aio_context (bs_), \ cond); }) -int bdrv_pdiscard (BdrvChild *child,int64_t offset,int bytes); -int bdrv_co_pdiscard (BdrvChild *child,int64_t offset,int bytes); +int bdrv_pdiscard (BdrvChild *child,int64_t offset,int64_t bytes); +int bdrv_co_pdiscard (BdrvChild *child,int64_t offset,int64_t bytes);int bdrv_has_zero_init_1 (BlockDriverState *bs);int bdrv_has_zero_init (BlockDriverState *bs);bool bdrv_unallocated_blocks_are_zero (BlockDriverState *bs);block/io.c +3
-2@@ -2653,7 +2653,7 @@ int bdrv_flush(BlockDriverState *bs)typedef struct DiscardCo {BdrvChild *child;int64_t offset; -int bytes; +int64_t bytes;int ret; }DiscardCo ; static void coroutine_fn bdrv_pdiscard_co_entry(void *opaque)@@ -2664,14 +2664,15 @@ static void coroutine_fn bdrv_pdiscard_co_entry(void *opaque)aio_wait_kick (); } -int coroutine_fn bdrv_co_pdiscard(BdrvChild *child, int64_t offset, int bytes) +int coroutine_fn bdrv_co_pdiscard(BdrvChild *child, int64_t offset, +int64_t bytes) {BdrvTrackedRequest req;int max_pdiscard, ret;int head, tail, align;BlockDriverState *bs = child->bs;Quoted 4 lines
block/io.c static void coroutine_fn bdrv_pdiscard_co_entry(void *opaque) · line 2675
Quoted 2 lines
block/io.c +1
-1-if (!bs || !bs->drv) { +if (!bs || !bs->drv || !bdrv_is_inserted (bs)) {Should we describe this change in the commit message?
Honestly, don't want to resend the series for this.
Quoted 3 lines
IIUC you added this check because you removed
bdrv_check_byte_request()below,
Quoted 7 lines
Quoted 2 lines
Maybe we can also remove '!bs->drv', since it is checked in
bdrv_is_inserted().Hmm, on v4 Kevin commented, that
bdrv_is_insertednot needed, and, as I understand, not only inbdrv_co_pdiscardit should be removed, but it may be done later. So, I'd prefer to keep it as is for now.
Make sense if it will be removed.
Quoted 19 lines
Quoted 8 lines
Quoted 8 lines
block/io.c +0
-0return -ENOMEDIUM ; }@@ -2679,9 +2680,8 @@ int coroutine_fn bdrv_co_pdiscard(BdrvChild *child, int64_t offset, int bytes)return -EPERM ; }Quoted 10 lines
block/io.c int coroutine_fn bdrv_co_pdiscard(BdrvChild *child, int64_t offset, int bytes) · line 2685
Quoted 6 lines
block/io.c +2
-3- ret =bdrv_check_byte_request (bs, offset, bytes); -if (ret <0 ) { -return ret; +if (offset <0 || bytes <0 || bytes >INT64_MAX - offset) { +return -EIO ; }Should we check if 'bytes' is greater than
'BDRV_REQUEST_MAX_SECTORS<<BDRV_SECTOR_BITS'?
Quoted 4 lines
No, as we are contrariwise trying to support large bytes parameter in
bdrv_co_pdiscard, which will exceed max request. If @bytes is large, it will be divided into several smaller requests in the following loop.
I understood.
I saw that we limit the request to the driver to 'max_pdiscard' or 'INT_MAX'.
As future work, could we use int64_t also for the driver callbacks?
Anyway, the patch LGTM.
Thanks,
Stefano
PATCH v5 3/3 iotests: test big qcow2 shrink
In reply to Vladimir Sementsov-OgievskiyThis test checks bug in qcow2_process_discards, fixed by previous commit.
--- tests/qemu-iotests/250 | 73 ++++++++++++++++++++++++++++++++++++++ tests/qemu-iotests/250.out | 23 ++++++++++++ tests/qemu-iotests/group | 1 + 3 files changed, 97 insertions(+) create mode 100755 tests/qemu-iotests/250 create mode 100644 tests/qemu-iotests/250.out
tests/qemu-iotests/250 new file +73 -0
@@ -0,0 +1,73 @@ +#!/usr/bin/env bash +# +# Test big discard in qcow2 shrink +# +# Copyright (c) 2019 Virtuozzo International GmbH. All rights reserved. +# +# This program is free software; you can redistribute it and/or modify +# it under the terms of the GNU General Public License as published by +# the Free Software Foundation; either version 2 of the License, or +# (at your option) any later version. +# +# This program is distributed in the hope that it will be useful, +# but WITHOUT ANY WARRANTY; without even the implied warranty of +# MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the +# GNU General Public License for more details. +# +# You should have received a copy of the GNU General Public License +# along with this program. If not, see <http://www.gnu.org/licenses/>. +# + +# creator [email protected] + +seq=`basename $0` +echo "QA output created by $seq" + +status=1 # failure is the default! + +_cleanup() +{ + _cleanup_test_img +} +trap "_cleanup; exit \$status" 0 1 2 3 15 + +# get standard environment, filters and checks +. ./common.rc +. ./common.filter + +_supported_fmt qcow2 +_supported_proto file +_supported_os Linux + +# This test checks that qcow2_process_discards does not truncate a discard +# request > 2G. +# To reproduce bug we need to overflow int by one sequential discard, so we +# need size > 2G, bigger cluster size (as with default 64k we may have maximum +# of 512M sequential data, corresponding to one L1 entry), and we need some +# data of the beginning of the disk mapped to the end of file to prevent +# bdrv_co_truncate(bs->file) call in qcow2_co_truncate(), which might succeed +# anyway. + +size=2100M +IMGOPTS="cluster_size=1M,preallocation=metadata" + +_make_test_img $size +$QEMU_IO -c 'discard 0 10M' -c 'discard 2090M 10M' \ + -c 'write 2090M 10M' -c 'write 0 10M' "$TEST_IMG" | _filter_qemu_io + +# Check that our trick with swapping first and last 10M chunks succeeded. +# Otherwise test may pass even if bdrv_pdiscard() fails in +# qcow2_process_discards() +$QEMU_IMG map "$TEST_IMG" | _filter_testdir + +$QEMU_IMG map -f raw "$TEST_IMG" | _filter_testdir + +$QEMU_IMG resize --shrink "$TEST_IMG" 5M + +$QEMU_IMG map -f raw "$TEST_IMG" | _filter_testdir + +# success, all done +echo "*** done" +rm -f $seq.full +status=0
tests/qemu-iotests/250.out new file +23 -0
@@ -0,0 +1,23 @@ +QA output created by 250 +Formatting 'TEST_DIR/t.IMGFMT', fmt=IMGFMT size=2202009600 preallocation=metadata +discard 10485760/10485760 bytes at offset 0 +10 MiB, X ops; XX:XX:XX.X (XXX YYY/sec and XXX ops/sec) +discard 10485760/10485760 bytes at offset 2191523840 +10 MiB, X ops; XX:XX:XX.X (XXX YYY/sec and XXX ops/sec) +wrote 10485760/10485760 bytes at offset 2191523840 +10 MiB, X ops; XX:XX:XX.X (XXX YYY/sec and XXX ops/sec) +wrote 10485760/10485760 bytes at offset 0 +10 MiB, X ops; XX:XX:XX.X (XXX YYY/sec and XXX ops/sec) +Offset Length Mapped to File +0 0xa00000 0x82f00000 TEST_DIR/t.qcow2 +0x82a00000 0xa00000 0x500000 TEST_DIR/t.qcow2 +Offset Length Mapped to File +0 0x301000 0 TEST_DIR/t.qcow2 +0x400000 0xb00000 0x400000 TEST_DIR/t.qcow2 +0x82f00000 0xa00000 0x82f00000 TEST_DIR/t.qcow2 +Image resized. +Offset Length Mapped to File +0 0x301000 0 TEST_DIR/t.qcow2 +0x400000 0x100000 0x400000 TEST_DIR/t.qcow2 +0x82f00000 0x500000 0x82f00000 TEST_DIR/t.qcow2 +*** done
tests/qemu-iotests/group +1 -0
@@ -248,3 +248,4 @@ 246 rw auto quick 247 rw auto quick 248 rw auto quick +250 rw auto quick
Signature · 1 line
2.18.0
ping
Quoted 20 lines · Vladimir Sementsov-Ogievskiy
v5: by Kevin's comments
02: add check for invalid @bytes parameter, drop r-b
03: move from qemu-img info to qemu-img map -f raw, drop r-b and t-bVladimir Sementsov-Ogievskiy (3):
block/qcow2-refcount: add trace-point toqcow2_process_discardsblock/io:bdrv_pdiscard: supportint64_tbytes parameter
iotests: test big qcow2 shrinkinclude/block/block.h | 4 +-- block/io.c | 16 ++++----- block/qcow2-refcount.c | 7 +++- block/trace-events | 3 ++ tests/qemu-iotests/250 | 73 ++++++++++++++++++++++++++++++++++++++ tests/qemu-iotests/250.out | 23 ++++++++++++ tests/qemu-iotests/group | 1 + 7 files changed, 116 insertions(+), 11 deletions(-)create mode 100755 tests/qemu-iotests/250
create mode 100644 tests/qemu-iotests/250.out
Signature · 2 lines
Best regards, Vladimir
[Qemu-devel] ping Re: [PATCH v5 0/3] Fix overflow bug in qcow2 discard
Quoted 25 lines · Vladimir Sementsov-Ogievskiy
ping
Quoted 20 lines · Vladimir Sementsov-Ogievskiy
v5: by Kevin's comments
02: add check for invalid @bytes parameter, drop r-b
03: move from qemu-img info to qemu-img map -f raw, drop r-b and t-bVladimir Sementsov-Ogievskiy (3):
block/qcow2-refcount: add trace-point toqcow2_process_discardsblock/io:bdrv_pdiscard: supportint64_tbytes parameter
iotests: test big qcow2 shrinkinclude/block/block.h | 4 +--
block/io.c | 16 ++++----- block/qcow2-refcount.c | 7 +++-
block/trace-events | 3 ++
tests/qemu-iotests/250 | 73 ++++++++++++++++++++++++++++++++++++++ tests/qemu-iotests/250.out | 23 ++++++++++++ tests/qemu-iotests/group | 1 +7 files changed, 116 insertions(+), 11 deletions(-)create mode 100755 tests/qemu-iotests/250
create mode 100644 tests/qemu-iotests/250.out
Signature · 2 lines
Best regards, Vladimir
Am 23.04.2019 um 14:57 hat Vladimir Sementsov-Ogievskiy geschrieben:
Quoted 3 lines
v5: by Kevin's comments
02: add check for invalid @bytes parameter, drop r-b
03: move from qemu-img info to qemu-img map -f raw, drop r-b and t-b
The patches look good to me, but the test case fails (this is on XFS; it does work on tmpfs):
home/kwolf/source/qemu/tests/qemu-iotests/250.out.bad 2019-06-03 15:37:37.202959101 +0200 renamed from home/kwolf/source/qemu/tests/qemu-iotests/250.out 2019-06-03 15:22:51.212961123 +0200 +2 -4
@@ -12,12 +12,10 @@ 0 0xa00000 0x82f00000 TEST_DIR/t.qcow2 0x82a00000 0xa00000 0x500000 TEST_DIR/t.qcow2 Offset Length Mapped to File -0 0x301000 0 TEST_DIR/t.qcow2 -0x400000 0xb00000 0x400000 TEST_DIR/t.qcow2 +0 0xf00000 0 TEST_DIR/t.qcow2 0x82f00000 0xa00000 0x82f00000 TEST_DIR/t.qcow2 Image resized. Offset Length Mapped to File -0 0x301000 0 TEST_DIR/t.qcow2 -0x400000 0x100000 0x400000 TEST_DIR/t.qcow2 +0 0x500000 0 TEST_DIR/t.qcow2 0x82f00000 0x500000 0x82f00000 TEST_DIR/t.qcow2 *** done
I'll apply the first two patches without the test for now, but please try if you can tweak the test case to work on all common filesystems.
Kevin
Quoted 26 lines · Kevin Wolf
Am 23.04.2019 um 14:57 hat Vladimir Sementsov-Ogievskiy geschrieben:
Quoted 3 lines
v5: by Kevin's comments
02: add check for invalid @bytes parameter, drop r-b
03: move from qemu-img info to qemu-img map -f raw, drop r-b and t-bThe patches look good to me, but the test case fails (this is on XFS; it does work on tmpfs):
home/kwolf/source/qemu/tests/qemu-iotests/250.out.bad 2019-06-03 15:37:37.202959101 +0200 renamed from home/kwolf/source/qemu/tests/qemu-iotests/250.out 2019-06-03 15:22:51.212961123 +0200 +2
-4@@ -12,12 +12,10 @@ 0 0xa00000 0x82f00000 TEST_DIR/t.qcow2 0x82a00000 0xa00000 0x500000 TEST_DIR/t.qcow2 Offset Length Mapped to File -0 0x301000 0 TEST_DIR/t.qcow2 -0x400000 0xb00000 0x400000 TEST_DIR/t.qcow2 +0 0xf00000 0 TEST_DIR/t.qcow2 0x82f00000 0xa00000 0x82f00000 TEST_DIR/t.qcow2 Image resized. Offset Length Mapped to File -0 0x301000 0 TEST_DIR/t.qcow2 -0x400000 0x100000 0x400000 TEST_DIR/t.qcow2 +0 0x500000 0 TEST_DIR/t.qcow2 0x82f00000 0x500000 0x82f00000 TEST_DIR/t.qcow2 *** done
Quoted 5 lines
I'll apply the first two patches without the test for now, but please try if you can tweak the test case to work on all common filesystems.
Kevin
Thank you! OK, I'll try
Signature · 2 lines
Best regards, Vladimir