Skip to contentSkip to conversation

fix & merge block_status_above and is_allocated_above

PATCH Superseded Vladimir Sementsov-Ogievskiy 2 participants 14 messages Sep 16, 2020 – Sep 24, 2020 Block layer

v6 of 7 Superseded Latest is v7 →

lore
Share & tools
lore
#1 lore raw

PATCH v6 1/5 block/io: fix bdrv_co_block_status_above

In reply to Vladimir Sementsov-Ogievskiy

bdrv_co_block_status_above has several design problems with handling short backing files:

1. With want_zeros=true, it may return ret with BDRV_BLOCK_ZERO but without BDRV_BLOCK_ALLOCATED flag, when actually short backing file which produces these after-EOF zeros is inside requested backing sequence.

2. With want_zero=false, it may return pnum=0 prior to actual EOF, because of EOF of short backing file.

Fix these things, making logic about short backing files clearer.

With fixed bdrv_block_status_above we also have to improve is_zero in qcow2 code, otherwise iotest 154 will fail, because with this patch we stop to merge zeros of different types (produced by fully unallocated in the whole backing chain regions vs produced by short backing files).

Note also, that this patch leaves for another day the general problem around block-status: misuse of BDRV_BLOCK_ALLOCATED as is-fs-allocated vs go-to-backing.

---
 block/io.c    | 66 ++++++++++++++++++++++++++++++++++++++++-----------
 block/qcow2.c | 16 +++++++++++--
 2 files changed, 66 insertions(+), 16 deletions(-)
block/io.c +52 -14
index 84f82bc069..e381d2da35 100644
@@ -2366,34 +2366,72 @@ bdrv_co_common_block_status_above(BlockDriverState *bs,
                                   int64_t *map,
                                   BlockDriverState **file)
 {
+    int ret;
     BlockDriverState *p;
-    int ret = 0;
-    bool first = true;
+    int64_t eof = 0;
 
     assert(bs != base);
-    for (p = bs; p != base; p = backing_bs(p)) {
+
+    ret = bdrv_co_block_status(bs, want_zero, offset, bytes, pnum, map, file);
+    if (ret < 0 || *pnum == 0 || ret & BDRV_BLOCK_ALLOCATED) {
+        return ret;
+    }
+
+    if (ret & BDRV_BLOCK_EOF) {
+        eof = offset + *pnum;
+    }
+
+    assert(*pnum <= bytes);
+    bytes = *pnum;
+
+    for (p = backing_bs(bs); p != base; p = backing_bs(p)) {
         ret = bdrv_co_block_status(p, want_zero, offset, bytes, pnum, map,
                                    file);
         if (ret < 0) {
-            break;
+            return ret;
         }
-        if (ret & BDRV_BLOCK_ZERO && ret & BDRV_BLOCK_EOF && !first) {
+        if (*pnum == 0) {
             /*
-             * Reading beyond the end of the file continues to read
-             * zeroes, but we can only widen the result to the
-             * unallocated length we learned from an earlier
-             * iteration.
+             * The top layer deferred to this layer, and because this layer is
+             * short, any zeroes that we synthesize beyond EOF behave as if they
+             * were allocated at this layer.
+             *
+             * We don't include BDRV_BLOCK_EOF into ret, as upper layer may be
+             * larger. We'll add BDRV_BLOCK_EOF if needed at function end, see
+             * below.
              */
+            assert(ret & BDRV_BLOCK_EOF);
             *pnum = bytes;
+            if (file) {
+                *file = p;
+            }
+            ret = BDRV_BLOCK_ZERO | BDRV_BLOCK_ALLOCATED;
+            break;
         }
-        if (ret & (BDRV_BLOCK_ZERO | BDRV_BLOCK_DATA)) {
+        if (ret & BDRV_BLOCK_ALLOCATED) {
+            /*
+             * We've found the node and the status, we must break.
+             *
+             * Drop BDRV_BLOCK_EOF, as it's not for upper layer, which may be
+             * larger. We'll add BDRV_BLOCK_EOF if needed at function end, see
+             * below.
+             */
+            ret &= ~BDRV_BLOCK_EOF;
             break;
         }
-        /* [offset, pnum] unallocated on this layer, which could be only
-         * the first part of [offset, bytes].  */
-        bytes = MIN(bytes, *pnum);
-        first = false;
+
+        /*
+         * OK, [offset, offset + *pnum) region is unallocated on this layer,
+         * let's continue the diving.
+         */
+        assert(*pnum <= bytes);
+        bytes = *pnum;
     }
+
+    if (offset + *pnum == eof) {
+        ret |= BDRV_BLOCK_EOF;
+    }
+
     return ret;
 }
 
block/qcow2.c +14 -2
index da56b1a4df..15ba0ce81a 100644
@@ -3872,8 +3872,20 @@ static bool is_zero(BlockDriverState *bs, int64_t offset, int64_t bytes)
     if (!bytes) {
         return true;
     }
-    res = bdrv_block_status_above(bs, NULL, offset, bytes, &nr, NULL, NULL);
-    return res >= 0 && (res & BDRV_BLOCK_ZERO) && nr == bytes;
+
+    /*
+     * bdrv_block_status_above doesn't merge different types of zeros, for
+     * example, zeros which come from the region which is unallocated in
+     * the whole backing chain, and zeros which comes because of a short
+     * backing file. So, we need a loop.
+     */
+    do {
+        res = bdrv_block_status_above(bs, NULL, offset, bytes, &nr, NULL, NULL);
+        offset += nr;
+        bytes -= nr;
+    } while (res >= 0 && (res & BDRV_BLOCK_ZERO) && nr && bytes);
+
+    return res >= 0 && (res & BDRV_BLOCK_ZERO) && bytes == 0;
 }
 
 static coroutine_fn int qcow2_co_pwrite_zeroes(BlockDriverState *bs,
Signature · 1 line
2.21.3
#3 lore raw

PATCH v6 2/5 block/io: bdrv_common_block_status_above: support include_base

In reply to Vladimir Sementsov-Ogievskiy

In order to reuse bdrv_common_block_status_above in bdrv_is_allocated_above, let's support include_base parameter.

---
 block/coroutines.h |  2 ++
 block/io.c         | 17 ++++++++++++-----
 2 files changed, 14 insertions(+), 5 deletions(-)
block/coroutines.h +2 -0
index f69179f5ef..1cb3128b94 100644
@@ -41,6 +41,7 @@ bdrv_pwritev(BdrvChild *child, int64_t offset, unsigned int bytes,
 int coroutine_fn
 bdrv_co_common_block_status_above(BlockDriverState *bs,
                                   BlockDriverState *base,
+                                  bool include_base,
                                   bool want_zero,
                                   int64_t offset,
                                   int64_t bytes,
@@ -50,6 +51,7 @@ bdrv_co_common_block_status_above(BlockDriverState *bs,
 int generated_co_wrapper
 bdrv_common_block_status_above(BlockDriverState *bs,
                                BlockDriverState *base,
+                               bool include_base,
                                bool want_zero,
                                int64_t offset,
                                int64_t bytes,
block/io.c +12 -5
index e381d2da35..0cc2dd7a3e 100644
@@ -2359,6 +2359,7 @@ early_out:
 int coroutine_fn
 bdrv_co_common_block_status_above(BlockDriverState *bs,
                                   BlockDriverState *base,
+                                  bool include_base,
                                   bool want_zero,
                                   int64_t offset,
                                   int64_t bytes,
@@ -2370,10 +2371,11 @@ bdrv_co_common_block_status_above(BlockDriverState *bs,
     BlockDriverState *p;
     int64_t eof = 0;
 
-    assert(bs != base);
+    assert(include_base || bs != base);
+    assert(!include_base || base); /* Can't include NULL base */
 
     ret = bdrv_co_block_status(bs, want_zero, offset, bytes, pnum, map, file);
-    if (ret < 0 || *pnum == 0 || ret & BDRV_BLOCK_ALLOCATED) {
+    if (ret < 0 || *pnum == 0 || ret & BDRV_BLOCK_ALLOCATED || bs == base) {
         return ret;
     }
 
@@ -2384,7 +2386,7 @@ bdrv_co_common_block_status_above(BlockDriverState *bs,
     assert(*pnum <= bytes);
     bytes = *pnum;
 
-    for (p = backing_bs(bs); p != base; p = backing_bs(p)) {
+    for (p = backing_bs(bs); include_base || p != base; p = backing_bs(p)) {
         ret = bdrv_co_block_status(p, want_zero, offset, bytes, pnum, map,
                                    file);
         if (ret < 0) {
@@ -2420,6 +2422,11 @@ bdrv_co_common_block_status_above(BlockDriverState *bs,
             break;
         }
 
+        if (p == base) {
+            assert(include_base);
+            break;
+        }
+
         /*
          * OK, [offset, offset + *pnum) region is unallocated on this layer,
          * let's continue the diving.
@@ -2439,7 +2446,7 @@ int bdrv_block_status_above(BlockDriverState *bs, BlockDriverState *base,
                             int64_t offset, int64_t bytes, int64_t *pnum,
                             int64_t *map, BlockDriverState **file)
 {
-    return bdrv_common_block_status_above(bs, base, true, offset, bytes,
+    return bdrv_common_block_status_above(bs, base, false, true, offset, bytes,
                                           pnum, map, file);
 }
 
@@ -2456,7 +2463,7 @@ int coroutine_fn bdrv_is_allocated(BlockDriverState *bs, int64_t offset,
     int ret;
     int64_t dummy;
 
-    ret = bdrv_common_block_status_above(bs, backing_bs(bs), false, offset,
+    ret = bdrv_common_block_status_above(bs, bs, true, false, offset,
                                          bytes, pnum ? pnum : &dummy, NULL,
                                          NULL);
     if (ret < 0) {
Signature · 1 line
2.21.3
#8 lore raw

PATCH v6 3/5 block/io: bdrv_common_block_status_above: support bs == base

In reply to Vladimir Sementsov-Ogievskiy

We are going to reuse bdrv_common_block_status_above in bdrv_is_allocated_above. bdrv_is_allocated_above may be called with include_base == false and still bs == base (for ex. from img_rebase()).

So, support this corner case.

---
 block/io.c | 6 +++++-
 1 file changed, 5 insertions(+), 1 deletion(-)
block/io.c +5 -1
index 0cc2dd7a3e..d864d035ac 100644
@@ -2371,9 +2371,13 @@ bdrv_co_common_block_status_above(BlockDriverState *bs,
     BlockDriverState *p;
     int64_t eof = 0;
 
-    assert(include_base || bs != base);
     assert(!include_base || base); /* Can't include NULL base */
 
+    if (!include_base && bs == base) {
+        *pnum = bytes;
+        return 0;
+    }
+
     ret = bdrv_co_block_status(bs, want_zero, offset, bytes, pnum, map, file);
     if (ret < 0 || *pnum == 0 || ret & BDRV_BLOCK_ALLOCATED || bs == base) {
         return ret;
Signature · 1 line
2.21.3
#0 lore raw

v6 cover fix & merge block_status_above and is_allocated_above

Hi all!

These series are here to address the following problem:
block-status-above functions may consider space after EOF of intermediate backing files as unallocated, which is wrong, as these backing files are the reason of producing zeroes, we never go further by backing chain after a short backing file. So, if such short-backing file is _inside_ requested sub-chain of the backing chain, we should never report space after its EOF as unallocated.

See patches 01,04,05 for details.

Note, that this series leaves for another day the general problem around block-status: misuse of BDRV_BLOCK_ALLOCATED as is-fs-allocated vs go-to-backing.
Audit for this problem is done here:
"backing chain & block status & filters"
https://lists.gnu.org/archive/html/qemu-devel/2020-04/msg04706.html
And I'm going to prepare series to address this problem.

Also, get_block_status func have same disease, but remains unfixed here:
I want to make separate series for it.

v6:

01: handle EOF better, don't merge reported ZERO-es with automatic after-EOF zeroes,
handle first layer out of the loop to make code read simpler. Drop r-b.
02: rebase on 01
05: update test output for extended l2 and backing file format, keep r-b

Based on series "PATCH v8 0/7] coroutines: generate wrapper code" or in other words:
Based-on: <[email protected]>

Vladimir Sementsov-Ogievskiy (5):
block/io: fix bdrv_co_block_status_above
block/io: bdrv_common_block_status_above: support include_base block/io: bdrv_common_block_status_above: support bs == base block/io: fix bdrv_is_allocated_above
iotests: add commit top->base cases to 274

 block/coroutines.h         |   2 +
 block/io.c                 | 126 +++++++++++++++++++++----------------
 block/qcow2.c              |  16 ++++-
 tests/qemu-iotests/274     |  20 ++++++
 tests/qemu-iotests/274.out |  68 ++++++++++++++++++++
 5 files changed, 175 insertions(+), 57 deletions(-)
Signature · 1 line
2.21.3
#10 lore raw

PATCH v6 4/5 block/io: fix bdrv_is_allocated_above

bdrv_is_allocated_above wrongly handles short backing files: it reports after-EOF space as UNALLOCATED which is wrong, as on read the data is generated on the level of short backing file (if all overlays has unallocated area at that place).

Reusing bdrv_common_block_status_above fixes the issue and unifies code path.

---
 block/io.c | 43 +++++--------------------------------------
 1 file changed, 5 insertions(+), 38 deletions(-)
block/io.c +5 -38
index d864d035ac..95b86429ca 100644
@@ -2491,52 +2491,19 @@ int coroutine_fn bdrv_is_allocated(BlockDriverState *bs, int64_t offset,
  * at 'offset + *pnum' may return the same allocation status (in other
  * words, the result is not necessarily the maximum possible range);
  * but 'pnum' will only be 0 when end of file is reached.
- *
  */
 int bdrv_is_allocated_above(BlockDriverState *top,
                             BlockDriverState *base,
                             bool include_base, int64_t offset,
                             int64_t bytes, int64_t *pnum)
 {
-    BlockDriverState *intermediate;
-    int ret;
-    int64_t n = bytes;
-
-    assert(base || !include_base);
-
-    intermediate = top;
-    while (include_base || intermediate != base) {
-        int64_t pnum_inter;
-        int64_t size_inter;
-
-        assert(intermediate);
-        ret = bdrv_is_allocated(intermediate, offset, bytes, &pnum_inter);
-        if (ret < 0) {
-            return ret;
-        }
-        if (ret) {
-            *pnum = pnum_inter;
-            return 1;
-        }
-
-        size_inter = bdrv_getlength(intermediate);
-        if (size_inter < 0) {
-            return size_inter;
-        }
-        if (n > pnum_inter &&
-            (intermediate == top || offset + pnum_inter < size_inter)) {
-            n = pnum_inter;
-        }
-
-        if (intermediate == base) {
-            break;
-        }
-
-        intermediate = backing_bs(intermediate);
+    int ret = bdrv_common_block_status_above(top, base, include_base, false,
+                                             offset, bytes, pnum, NULL, NULL);
+    if (ret < 0) {
+        return ret;
     }
 
-    *pnum = n;
-    return 0;
+    return !!(ret & BDRV_BLOCK_ALLOCATED);
 }
 
 int coroutine_fn
Signature · 1 line
2.21.3
#12 lore raw

PATCH v6 5/5 iotests: add commit top->base cases to 274

In reply to Vladimir Sementsov-Ogievskiy

These cases are fixed by previous patches around block_status and is_allocated.

---
 tests/qemu-iotests/274     | 20 +++++++++++
 tests/qemu-iotests/274.out | 68 ++++++++++++++++++++++++++++++++++++++
 2 files changed, 88 insertions(+)
tests/qemu-iotests/274 +20 -0
index d4571c5465..76b1ba6a52 100755
@@ -115,6 +115,26 @@ with iotests.FilePath('base') as base, \
     iotests.qemu_io_log('-c', 'read -P 1 0 %d' % size_short, mid)
     iotests.qemu_io_log('-c', 'read -P 0 %d %d' % (size_short, size_diff), mid)
 
+    iotests.log('=== Testing qemu-img commit (top -> base) ===')
+
+    create_chain()
+    iotests.qemu_img_log('commit', '-b', base, top)
+    iotests.img_info_log(base)
+    iotests.qemu_io_log('-c', 'read -P 1 0 %d' % size_short, base)
+    iotests.qemu_io_log('-c', 'read -P 0 %d %d' % (size_short, size_diff), base)
+
+    iotests.log('=== Testing QMP active commit (top -> base) ===')
+
+    create_chain()
+    with create_vm() as vm:
+        vm.launch()
+        vm.qmp_log('block-commit', device='top', base_node='base',
+                   job_id='job0', auto_dismiss=False)
+        vm.run_job('job0', wait=5)
+
+    iotests.img_info_log(mid)
+    iotests.qemu_io_log('-c', 'read -P 1 0 %d' % size_short, base)
+    iotests.qemu_io_log('-c', 'read -P 0 %d %d' % (size_short, size_diff), base)
 
     iotests.log('== Resize tests ==')
 
tests/qemu-iotests/274.out +68 -0

74 lines not on this page · raw message

Signature · 1 line
2.21.3
#4 lore raw

Re: [PATCH v6 2/5] block/io: bdrv_common_block_status_above: support include_base

In reply to Vladimir Sementsov-Ogievskiy
Quoted 6 lines · Vladimir Sementsov-Ogievskiy
block/io.c +1 -1
-    for (p = backing_bs(bs); p != base; p = backing_bs(p)) {
+    for (p = backing_bs(bs); include_base || p != base; p = backing_bs(p)) {
         ret = bdrv_co_block_status(p, want_zero, offset, bytes, pnum, map,
                                    file);
         if (ret < 0) {

block/io.c bdrv_co_common_block_status_above(BlockDriverState *bs, · line 2429

Quoted 8 lines
block/io.c +5 -0
@@ -2420,6 +2422,11 @@ bdrv_co_common_block_status_above(BlockDriverState *bs,
             break;
         }
 
+        if (p == base) {
+            assert(include_base);
+            break;
+        }
+

Another option is something like:

   BlockDriverState *last_bs = include_base ? base : backing_bs(base);

and you get a simpler 'for' loop.

But why do we need include_base at all? Can't the caller just pass backing_bs(base) instead? I'm talking also about the existing case of bdrv_is_allocated_above().

Berto

Reviewed-by on v6 3/5
#9 lore raw

Re: [PATCH v6 3/5] block/io: bdrv_common_block_status_above: support bs == base

In reply to Vladimir Sementsov-Ogievskiy
Quoted 9 lines · Vladimir Sementsov-Ogievskiy

We are going to reuse bdrv_common_block_status_above in bdrv_is_allocated_above. bdrv_is_allocated_above may be called with include_base == false and still bs == base (for ex. from img_rebase()).

So, support this corner case.

(my question about include_base remains, but otherwise this patch is correct)

Berto

#5 lore raw

Re: [PATCH v6 2/5] block/io: bdrv_common_block_status_above: support include_base

In reply to Alberto Garcia
Quoted 18 lines · Alberto Garcia
Quoted 6 lines · Vladimir Sementsov-Ogievskiy
block/io.c +1 -1
-    for (p = backing_bs(bs); p != base; p = backing_bs(p)) {
+    for (p = backing_bs(bs); include_base || p != base; p = backing_bs(p)) {
          ret = bdrv_co_block_status(p, want_zero, offset, bytes, pnum, map,
                                     file);
          if (ret < 0) {

block/io.c bdrv_co_common_block_status_above(BlockDriverState *bs, · line 2429

Quoted 8 lines
block/io.c +5 -0
@@ -2420,6 +2422,11 @@ bdrv_co_common_block_status_above(BlockDriverState *bs,
              break;
          }
 
+        if (p == base) {
+            assert(include_base);
+            break;
+        }
+

Another option is something like:

Quoted 1 line
    BlockDriverState *last_bs = include_base ? base : backing_bs(base);

hmm, in case when include_base is false, last bs is not backing_bs(base) but the parent of base.

Quoted 7 lines

and you get a simpler 'for' loop.

But why do we need include_base at all? Can't the caller just pass backing_bs(base) instead? I'm talking also about the existing case of bdrv_is_allocated_above().

include_base was introduced for the case when caller doesn't own backing_bs(base), and therefore shouldn't do operations that may yield (block_status can) dependent on backing_bs(base). In particular, in block stream, where link to base is not frozen.

Signature · 2 lines
Best regards,
Vladimir
#6 lore raw

Re: [PATCH v6 2/5] block/io: bdrv_common_block_status_above: support include_base

In reply to Vladimir Sementsov-Ogievskiy
Quoted 4 lines · Vladimir Sementsov-Ogievskiy
Quoted 1 line
    BlockDriverState *last_bs = include_base ? base : backing_bs(base);

hmm, in case when include_base is false, last bs is not backing_bs(base) but the parent of base.

Oops, yes, it should be the other way around %-)

Quoted 8 lines
Quoted 3 lines

But why do we need include_base at all? Can't the caller just pass backing_bs(base) instead? I'm talking also about the existing case of bdrv_is_allocated_above().

include_base was introduced for the case when caller doesn't own backing_bs(base), and therefore shouldn't do operations that may yield (block_status can) dependent on backing_bs(base). In particular, in block stream, where link to base is not frozen.

You're right, thanks!

Berto

Reviewed-by on v6 1/5
#2 lore raw

Re: [PATCH v6 1/5] block/io: fix bdrv_co_block_status_above

In reply to Vladimir Sementsov-Ogievskiy
Quoted 18 lines · Vladimir Sementsov-Ogievskiy

bdrv_co_block_status_above has several design problems with handling short backing files:

1. With want_zeros=true, it may return ret with BDRV_BLOCK_ZERO but without BDRV_BLOCK_ALLOCATED flag, when actually short backing file which produces these after-EOF zeros is inside requested backing sequence.

2. With want_zero=false, it may return pnum=0 prior to actual EOF, because of EOF of short backing file.

Fix these things, making logic about short backing files clearer.

With fixed bdrv_block_status_above we also have to improve is_zero in qcow2 code, otherwise iotest 154 will fail, because with this patch we stop to merge zeros of different types (produced by fully unallocated in the whole backing chain regions vs produced by short backing files).

Quoted 5 lines

Note also, that this patch leaves for another day the general problem around block-status: misuse of BDRV_BLOCK_ALLOCATED as is-fs-allocated vs go-to-backing.

Berto

Reviewed-by on v6 4/5
#11 lore raw

Re: [PATCH v6 4/5] block/io: fix bdrv_is_allocated_above

In reply to Vladimir Sementsov-Ogievskiy
Quoted 10 lines · Vladimir Sementsov-Ogievskiy

bdrv_is_allocated_above wrongly handles short backing files: it reports after-EOF space as UNALLOCATED which is wrong, as on read the data is generated on the level of short backing file (if all overlays has unallocated area at that place).

Reusing bdrv_common_block_status_above fixes the issue and unifies code path.

Berto