block: Split padded I/O vectors exceeding IOV_MAX
v2 cover block: Split padded I/O vectors exceeding IOV_MAX
RFC:
https://lists.nongnu.org/archive/html/qemu-block/2023-03/msg00446.html
v1:
https://lists.nongnu.org/archive/html/qemu-devel/2023-03/msg05049.html
As explained in the RFC’s cover letter, the problem this series addresses is that we pad requests from the guest that are unaligned to the underlying storage’s alignment requirements so they fit. This involves giving them head and/or tail padding.
We generally work with I/O vectors, so this padding is added via prepending/appending vector elements. We have a maximum limit on the number of elements, though (1024, specifically), so it is possible that padding has the vector exceed this limit, resulting in an unrecoverable I/O error that is returned to the guest -- on a perfectly valid request, as far as the guest is concerned.
To fix this, when the limit would be exceeded, this series temporarily merges some (up to three) I/O vector element into one to decrease the number of vector elements as far as necessary.
v2:
- Patch 1: Made a note in the commit message of having renamed
qiov_slice() -> qemu_iovec_slice()
- Patch 2:
- Renamed bdrv_padding_destroy() to bdrv_padding_finalize(), indicating that the padding is not just simply destroyed, but more steps may be taken (i.e. copying back the contents of the temporary buffer used for the merged elements)
- Generally replaced “couple of” by “two or three”
Hanna Czenczek (4):
util/iov: Make qiov_slice() public
block: Collapse padded I/O vecs exceeding IOV_MAX util/iov: Remove qemu_iovec_init_extended() iotests/iov-padding: New test
include/qemu/iov.h | 8 +- block/io.c | 166 +++++++++++++++++++++-- util/iov.c | 89 +++--------- tests/qemu-iotests/tests/iov-padding | 85 ++++++++++++ tests/qemu-iotests/tests/iov-padding.out | 59 ++++++++ 5 files changed, 314 insertions(+), 93 deletions(-) create mode 100755 tests/qemu-iotests/tests/iov-padding create mode 100644 tests/qemu-iotests/tests/iov-padding.out
Signature · 1 line
2.39.1
PATCH v2 1/4 util/iov: Make qiov_slice() public
We want to inline qemu_iovec_init_extended() in block/io.c for padding requests, and having access to qiov_slice() is useful for this. As a public function, it is renamed to qemu_iovec_slice().
(We will need to count the number of I/O vector elements of a slice there, and then later process this slice. Without qiov_slice(), we would need to call qemu_iovec_subvec_niov(), and all further IOV-processing functions may need to skip prefixing elements to accomodate for a qiov_offset. Because qemu_iovec_subvec_niov() internally calls qiov_slice(), we can just have the block/io.c code call qiov_slice() itself, thus get the number of elements, and also create an iovec array with the superfluous prefixing elements stripped, so the following processing functions no longer need to skip them.)
--- include/qemu/iov.h | 3 +++ util/iov.c | 14 +++++++------- 2 files changed, 10 insertions(+), 7 deletions(-)
include/qemu/iov.h +3 -0
@@ -229,6 +229,9 @@ int qemu_iovec_init_extended( void *tail_buf, size_t tail_len);void qemu_iovec_init_slice (QEMUIOVector *qiov,QEMUIOVector *source,size_t offset,size_t len); +struct iovec *qemu_iovec_slice (QEMUIOVector *qiov, +size_t offset,size_t len, +size_t *head,size_t *tail,int *niov);int qemu_iovec_subvec_niov (QEMUIOVector *qiov,size_t offset,size_t len);void qemu_iovec_add (QEMUIOVector *qiov,void *base,size_t len); void qemu_iovec_concat(QEMUIOVector *dst,
util/iov.c +7 -7
@@ -378,15 +378,15 @@ static struct iovec *iov_skip_offset(struct iovec *iov, size_t offset, }/* -* qiov_slice +* qemu_iovec_slice * * Find subarray of iovec's, containing requested range. @head would * be offset in first iov (returned by the function), @tail would be * count of extra bytes in last iovec (returned iov + @niov - 1). */ -static struct iovec *qiov_slice (QEMUIOVector *qiov, -size_t offset,size_t len, -size_t *head,size_t *tail,int *niov) +struct iovec *qemu_iovec_slice (QEMUIOVector *qiov, +size_t offset,size_t len, +size_t *head,size_t *tail,int *niov) {struct iovec *iov, *end_iov;
@@ -411,7 +411,7 @@ int qemu_iovec_subvec_niov(QEMUIOVector *qiov, size_t offset, size_t len)size_t head, tail;int niov; -qiov_slice (qiov, offset, len, &head, &tail, &niov); +qemu_iovec_slice (qiov, offset, len, &head, &tail, &niov);return niov; }
@@ -439,8 +439,8 @@ int qemu_iovec_init_extended( }if (mid_len) { - mid_iov =qiov_slice (mid_qiov, mid_offset, mid_len, - &mid_head, &mid_tail, &mid_niov); + mid_iov =qemu_iovec_slice (mid_qiov, mid_offset, mid_len, + &mid_head, &mid_tail, &mid_niov); } total_niov = !!head_len + mid_niov + !!tail_len;
Signature · 1 line
2.39.1
PATCH v2 2/4 block: Collapse padded I/O vecs exceeding IOV_MAX
In reply to Hanna CzenczekWhen processing vectored guest requests that are not aligned to the storage request alignment, we pad them by adding head and/or tail buffers for a read-modify-write cycle.
The guest can submit I/O vectors up to IOV_MAX (1024) in length, but with this padding, the vector can exceed that limit. As of 4c002cef0e9abe7135d7916c51abce47f7fc1ee2 ("util/iov: make qemu_iovec_init_extended() honest"), we refuse to pad vectors beyond the limit, instead returning an error to the guest.
To the guest, this appears as a random I/O error. We should not return an I/O error to the guest when it issued a perfectly valid request.
Before 4c002cef0e9abe7135d7916c51abce47f7fc1ee2, we just made the vector longer than IOV_MAX, which generally seems to work (because the guest assumes a smaller alignment than we really have, file-posix's raw_co_prw() will generally see bdrv_qiov_is_aligned() return false, and so emulate the request, so that the IOV_MAX does not matter). However, that does not seem exactly great.
I see two ways to fix this problem:
1. We split such long requests into two requests.
2. We join some elements of the vector into new buffers to make it shorter.
I am wary of (1), because it seems like it may have unintended side effects.
(2) on the other hand seems relatively simple to implement, with hopefully few side effects, so this patch does that.
To do this, the use of qemu_iovec_init_extended() in bdrv_pad_request() is effectively replaced by the new function bdrv_create_padded_qiov(), which not only wraps the request IOV with padding head/tail, but also ensures that the resulting vector will not have more than IOV_MAX elements. Putting that functionality into qemu_iovec_init_extended() is infeasible because it requires allocating a bounce buffer; doing so would require many more parameters (buffer alignment, how to initialize the buffer, and out parameters like the buffer, its length, and the original elements), which is not reasonable.
Conversely, it is not difficult to move qemu_iovec_init_extended()'s functionality into bdrv_create_padded_qiov() by using public qemu_iovec_* functions, so that is what this patch does.
Because bdrv_pad_request() was the only "serious" user of qemu_iovec_init_extended(), the next patch will remove the latter function, so the functionality is not implemented twice.
--- block/io.c | 166 ++++++++++++++++++++++++++++++++++++++++++++++++----- 1 file changed, 151 insertions(+), 15 deletions(-)
block/io.c +151 -15
@@ -1435,6 +1435,14 @@ out:* @merge_reads is true for small requests, * if @buf_len == @head + bytes + @tail. In this case it is possible that both * head and tail exist but @buf_len == align and @tail_buf == @buf. +* +* @write is true for write requests, false for read requests. +* +* If padding makes the vector too long (exceeding IOV_MAX), then we need to +* merge existing vector elements into a single one. @collapse_bounce_buf acts +* as the bounce buffer in such cases. @pre_collapse_qiov has the pre-collapse +* I/O vector elements so for read requests, the data can be copied back after +* the read is done. */ typedef struct BdrvRequestPadding { uint8_t *buf;
@@ -1443,11 +1451,17 @@ typedef struct BdrvRequestPadding {size_t head;size_t tail;bool merge_reads; +bool write;QEMUIOVector local_qiov; + +uint8_t *collapse_bounce_buf; +size_t collapse_len; +QEMUIOVector pre_collapse_qiov; } BdrvRequestPadding;static bool bdrv_init_padding (BlockDriverState *bs,int64_t offset,int64_t bytes, +bool write,BdrvRequestPadding *pad) {int64_t align = bs->bl.request_alignment;
@@ -1479,6 +1493,8 @@ static bool bdrv_init_padding(BlockDriverState *bs, pad->tail_buf = pad->buf + pad->buf_len - align; } + pad->write = write; +return true; }
@@ -1543,8 +1559,23 @@ zero_mem:return 0 ; } -static void bdrv_padding_destroy (BdrvRequestPadding *pad) +/** +* Free *pad's associated buffers, and perform any necessary finalization steps. +*/ +static void bdrv_padding_finalize (BdrvRequestPadding *pad) { +if (pad->collapse_bounce_buf) { +if (!pad->write) { +/* +* If padding required elements in the vector to be collapsed into a +* bounce buffer, copy the bounce buffer content back +*/ +qemu_iovec_from_buf (&pad->pre_collapse_qiov,0 , + pad->collapse_bounce_buf, pad->collapse_len); + } +qemu_vfree (pad->collapse_bounce_buf); +qemu_iovec_destroy (&pad->pre_collapse_qiov); + }if (pad->buf) {qemu_vfree (pad->buf);qemu_iovec_destroy (&pad->local_qiov);
@@ -1552,6 +1583,101 @@ static void bdrv_padding_destroy(BdrvRequestPadding *pad)memset (pad,0 ,sizeof (*pad)); } +/* +* Create pad->local_qiov by wrapping @iov in the padding head and tail, while +* ensuring that the resulting vector will not exceed IOV_MAX elements. +* +* To ensure this, when necessary, the first two or three elements of @iov are +* merged into pad->collapse_bounce_buf and replaced by a reference to that +* bounce buffer in pad->local_qiov. +* +* After performing a read request, the data from the bounce buffer must be +* copied back into pad->pre_collapse_qiov (e.g. by bdrv_padding_finalize()). +*/ +static int bdrv_create_padded_qiov (BlockDriverState *bs, +BdrvRequestPadding *pad, +struct iovec *iov,int niov, +size_t iov_offset,size_t bytes) +{ +int padded_niov, surplus_count, collapse_count; + +/* Assert this invariant */ +assert (niov <=IOV_MAX ); + +/* +* Cannot pad if resulting length would exceed SIZE_MAX. Returning an error +* to the guest is not ideal, but there is little else we can do. At least +* this will practically never happen on 64-bit systems. +*/ +if (SIZE_MAX - pad->head < bytes || +SIZE_MAX - pad->head - bytes < pad->tail) + { +return -EINVAL ; + } + +/* Length of the resulting IOV if we just concatenated everything */ + padded_niov = !!pad->head + niov + !!pad->tail; + +qemu_iovec_init (&pad->local_qiov,MIN (padded_niov,IOV_MAX )); + +if (pad->head) { +qemu_iovec_add (&pad->local_qiov, pad->buf, pad->head); + } + +/* +* If padded_niov > IOV_MAX, we cannot just concatenate everything. +* Instead, merge the first two or three elements of @iov to reduce the +* number of vector elements as necessary. +*/ +if (padded_niov >IOV_MAX ) { +/* +* Only head and tail can have lead to the number of entries exceeding +* IOV_MAX, so we can exceed it by the head and tail at most. We need +* to reduce the number of elements by `surplus_count`, so we merge that +* many elements plus one into one element. +*/ + surplus_count = padded_niov -IOV_MAX ; +assert (surplus_count <= !!pad->head + !!pad->tail); + collapse_count = surplus_count +1 ; + +/* +* Move the elements to collapse into `pad->pre_collapse_qiov`, then +* advance `iov` (and associated variables) by those elements. +*/ +qemu_iovec_init (&pad->pre_collapse_qiov, collapse_count); +qemu_iovec_concat_iov (&pad->pre_collapse_qiov, iov, + collapse_count, iov_offset,SIZE_MAX ); + iov += collapse_count; + iov_offset =0 ; + niov -= collapse_count; + bytes -= pad->pre_collapse_qiov.size; + +/* +* Construct the bounce buffer to match the length of the to-collapse +* vector elements, and for write requests, initialize it with the data +* from those elements. Then add it to `pad->local_qiov`. +*/ + pad->collapse_len = pad->pre_collapse_qiov.size; + pad->collapse_bounce_buf =qemu_blockalign (bs, pad->collapse_len); +if (pad->write) { +qemu_iovec_to_buf (&pad->pre_collapse_qiov,0 , + pad->collapse_bounce_buf, pad->collapse_len); + } +qemu_iovec_add (&pad->local_qiov, + pad->collapse_bounce_buf, pad->collapse_len); + } + +qemu_iovec_concat_iov (&pad->local_qiov, iov, niov, iov_offset, bytes); + +if (pad->tail) { +qemu_iovec_add (&pad->local_qiov, + pad->buf + pad->buf_len - pad->tail, pad->tail); + } + +assert (pad->local_qiov.niov ==MIN (padded_niov,IOV_MAX )); +return 0 ; +} +/* * bdrv_pad_request *
@@ -1559,6 +1685,8 @@ static void bdrv_padding_destroy(BdrvRequestPadding *pad)* read of padding, bdrv_padding_rmw_read() should be called separately if * needed. * +* @write is true for write requests, false for read requests. +* * Request parameters (@qiov, &qiov_offset, &offset, &bytes) are in-out: * - on function start they represent original request * - on failure or when padding is not needed they are unchanged
@@ -1567,26 +1695,34 @@ static void bdrv_padding_destroy(BdrvRequestPadding *pad)static int bdrv_pad_request (BlockDriverState *bs,QEMUIOVector **qiov,size_t *qiov_offset,int64_t *offset,int64_t *bytes, +bool write,BdrvRequestPadding *pad,bool *padded,BdrvRequestFlags *flags) {int ret; +struct iovec *sliced_iov; +int sliced_niov; +size_t sliced_head, sliced_tail;bdrv_check_qiov_request (*offset, *bytes, *qiov, *qiov_offset, &error_abort); -if (!bdrv_init_padding (bs, *offset, *bytes, pad)) { +if (!bdrv_init_padding (bs, *offset, *bytes, write, pad)) {if (padded) { *padded = false; }return 0 ; } - ret =qemu_iovec_init_extended (&pad->local_qiov, pad->buf, pad->head, - *qiov, *qiov_offset, *bytes, - pad->buf + pad->buf_len - pad->tail, - pad->tail); + sliced_iov =qemu_iovec_slice (*qiov, *qiov_offset, *bytes, + &sliced_head, &sliced_tail, + &sliced_niov); + +/* Guaranteed by bdrv_check_qiov_request() */ +assert (*bytes <=SIZE_MAX ); + ret =bdrv_create_padded_qiov (bs, pad, sliced_iov, sliced_niov, + sliced_head, *bytes);if (ret <0 ) { -bdrv_padding_destroy (pad); +bdrv_padding_finalize (pad);return ret; } *bytes += pad->head + pad->tail;
@@ -1653,8 +1789,8 @@ int coroutine_fn bdrv_co_preadv_part(BdrvChild *child, flags |=BDRV_REQ_COPY_ON_READ ; } - ret =bdrv_pad_request (bs, &qiov, &qiov_offset, &offset, &bytes, &pad, -NULL , &flags); + ret =bdrv_pad_request (bs, &qiov, &qiov_offset, &offset, &bytes, false, + &pad,NULL , &flags);if (ret <0 ) { goto fail; }
@@ -1664,7 +1800,7 @@ int coroutine_fn bdrv_co_preadv_part(BdrvChild *child, bs->bl.request_alignment, qiov, qiov_offset, flags);tracked_request_end (&req); -bdrv_padding_destroy (&pad); +bdrv_padding_finalize (&pad); fail:bdrv_dec_in_flight (bs);
@@ -1996,7 +2132,7 @@ bdrv_co_do_zero_pwritev(BdrvChild *child, int64_t offset, int64_t bytes,/* This flag doesn't make sense for padding or zero writes */ flags &= ~BDRV_REQ_REGISTERED_BUF ; - padding =bdrv_init_padding (bs, offset, bytes, &pad); + padding =bdrv_init_padding (bs, offset, bytes, true, &pad);if (padding) {assert (!(flags &BDRV_REQ_NO_WAIT ));bdrv_make_request_serialising (req, align);
@@ -2044,7 +2180,7 @@ bdrv_co_do_zero_pwritev(BdrvChild *child, int64_t offset, int64_t bytes, } out: -bdrv_padding_destroy (&pad); +bdrv_padding_finalize (&pad);return ret; }
@@ -2112,8 +2248,8 @@ int coroutine_fn bdrv_co_pwritev_part(BdrvChild *child,* bdrv_co_do_zero_pwritev() does aligning by itself, so, we do * alignment only if there is no ZERO flag. */ - ret =bdrv_pad_request (bs, &qiov, &qiov_offset, &offset, &bytes, &pad, - &padded, &flags); + ret =bdrv_pad_request (bs, &qiov, &qiov_offset, &offset, &bytes, true, + &pad, &padded, &flags);if (ret <0 ) {return ret; }
@@ -2143,7 +2279,7 @@ int coroutine_fn bdrv_co_pwritev_part(BdrvChild *child, ret =bdrv_aligned_pwritev (child, &req, offset, bytes, align, qiov, qiov_offset, flags); -bdrv_padding_destroy (&pad); +bdrv_padding_finalize (&pad); out:tracked_request_end (&req);
Signature · 1 line
2.39.1
Re: [PATCH v2 2/4] block: Collapse padded I/O vecs exceeding IOV_MAX
Quoted 45 lines · Hanna Czenczek
When processing vectored guest requests that are not aligned to the storage request alignment, we pad them by adding head and/or tail buffers for a read-modify-write cycle.
The guest can submit I/O vectors up to
IOV_MAX(1024) in length, but with this padding, the vector can exceed that limit. As of4c002cef0e9abe7135d7916c51abce47f7fc1ee2("util/iov: makeqemu_iovec_init_extended()honest"), we refuse to pad vectors beyond the limit, instead returning an error to the guest.To the guest, this appears as a random I/O error. We should not return an I/O error to the guest when it issued a perfectly valid request.
Before
4c002cef0e9abe7135d7916c51abce47f7fc1ee2, we just made the vector longer thanIOV_MAX, which generally seems to work (because the guest assumes a smaller alignment than we really have, file-posix'sraw_co_prw()will generally seebdrv_qiov_is_aligned()return false, and so emulate the request, so that theIOV_MAXdoes not matter). However, that does not seem exactly great.I see two ways to fix this problem:
1. We split such long requests into two requests.
2. We join some elements of the vector into new buffers to make it shorter.I am wary of (1), because it seems like it may have unintended side effects.
(2) on the other hand seems relatively simple to implement, with hopefully few side effects, so this patch does that.
To do this, the use of
qemu_iovec_init_extended()inbdrv_pad_request()is effectively replaced by the new functionbdrv_create_padded_qiov(), which not only wraps the request IOV with padding head/tail, but also ensures that the resulting vector will not have more thanIOV_MAXelements. Putting that functionality intoqemu_iovec_init_extended()is infeasible because it requires allocating a bounce buffer; doing so would require many more parameters (buffer alignment, how to initialize the buffer, and out parameters like the buffer, its length, and the original elements), which is not reasonable.Conversely, it is not difficult to move qemu_iovec_init_extended()'s functionality into
bdrv_create_padded_qiov()by using public qemu_iovec_* functions, so that is what this patch does.
Quoted 6 lines
Because
bdrv_pad_request()was the only "serious" user ofqemu_iovec_init_extended(), the next patch will remove the latter function, so the functionality is not implemented twice.Signed-off-by Hanna Czenczek<[email protected]>
Signature · 2 lines
Best regards, Vladimir
PATCH v2 3/4 util/iov: Remove qemu_iovec_init_extended()
In reply to Hanna Czenczekbdrv_pad_request() was the main user of qemu_iovec_init_extended(). HEAD^ has removed that use, so we can remove qemu_iovec_init_extended() now.
The only remaining user is qemu_iovec_init_slice(), which can easily inline the small part it really needs.
Note that qemu_iovec_init_extended() offered a memcpy() optimization to initialize the new I/O vector. qemu_iovec_concat_iov(), which is used to replace its functionality, does not, but calls qemu_iovec_add() for every single element. If we decide this optimization was important, we will need to re-implement it in qemu_iovec_concat_iov(), which might also benefit its pre-existing users.
--- include/qemu/iov.h | 5 --- util/iov.c | 79 +++++++--------------------------------------- 2 files changed, 11 insertions(+), 73 deletions(-)
include/qemu/iov.h +0 -5
@@ -222,11 +222,6 @@ static inline void *qemu_iovec_buf(QEMUIOVector *qiov)void qemu_iovec_init (QEMUIOVector *qiov,int alloc_hint);void qemu_iovec_init_external (QEMUIOVector *qiov,struct iovec *iov,int niov); -int qemu_iovec_init_extended ( -QEMUIOVector *qiov, -void *head_buf,size_t head_len, -QEMUIOVector *mid_qiov,size_t mid_offset,size_t mid_len, -void *tail_buf,size_t tail_len);void qemu_iovec_init_slice (QEMUIOVector *qiov,QEMUIOVector *source,size_t offset,size_t len); struct iovec *qemu_iovec_slice(QEMUIOVector *qiov,
util/iov.c +11 -68
95 lines not on this page · raw message
Signature · 1 line
2.39.1
PATCH v2 4/4 iotests/iov-padding: New test
In reply to Hanna CzenczekTest that even vectored IO requests with 1024 vector elements that are not aligned to the device's request alignment will succeed.
--- tests/qemu-iotests/tests/iov-padding | 85 ++++++++++++++++++++++++ tests/qemu-iotests/tests/iov-padding.out | 59 ++++++++++++++++ 2 files changed, 144 insertions(+) create mode 100755 tests/qemu-iotests/tests/iov-padding create mode 100644 tests/qemu-iotests/tests/iov-padding.out
tests/qemu-iotests/tests/iov-padding new file +85 -0
@@ -0,0 +1,85 @@ +#!/usr/bin/env bash +# group: rw quick +# +# Check the interaction of request padding (to fit alignment restrictions) with +# vectored I/O from the guest +# +# Copyright Red Hat +# +# 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/>. +# + +seq=$(basename $0) +echo "QA output created by $seq" + +status=1 # failure is the default! + +_cleanup() +{ + _cleanup_test_img +} +trap "_cleanup; exit \$status" 0 1 2 3 15 + +# get standard environment, filters and checks +cd .. +. ./common.rc +. ./common.filter + +_supported_fmt raw +_supported_proto file + +_make_test_img 1M + +IMGSPEC="driver=blkdebug,align=4096,image.driver=file,image.filename=$TEST_IMG" + +# Four combinations: +# - Offset 4096, length 1023 * 512 + 512: Fully aligned to 4k +# - Offset 4096, length 1023 * 512 + 4096: Head is aligned, tail is not +# - Offset 512, length 1023 * 512 + 512: Neither head nor tail are aligned +# - Offset 512, length 1023 * 512 + 4096: Tail is aligned, head is not +for start_offset in 4096 512; do + for last_element_length in 512 4096; do + length=$((1023 * 512 + $last_element_length)) + + echo + echo "== performing 1024-element vectored requests to image (offset: $start_offset; length: $length) ==" + + # Fill with data for testing + $QEMU_IO -c 'write -P 1 0 1M' "$TEST_IMG" | _filter_qemu_io + + # 1023 512-byte buffers, and then one with length $last_element_length + cmd_params="-P 2 $start_offset $(yes 512 | head -n 1023 | tr '\n' ' ') $last_element_length" + QEMU_IO_OPTIONS="$QEMU_IO_OPTIONS_NO_FMT" $QEMU_IO \ + -c "writev $cmd_params" \ + --image-opts \ + "$IMGSPEC" \ + | _filter_qemu_io + + # Read all patterns -- read the part we just wrote with writev twice, + # once "normally", and once with a readv, so we see that that works, too + QEMU_IO_OPTIONS="$QEMU_IO_OPTIONS_NO_FMT" $QEMU_IO \ + -c "read -P 1 0 $start_offset" \ + -c "read -P 2 $start_offset $length" \ + -c "readv $cmd_params" \ + -c "read -P 1 $((start_offset + length)) $((1024 * 1024 - length - start_offset))" \ + --image-opts \ + "$IMGSPEC" \ + | _filter_qemu_io + done +done + +# success, all done +echo "*** done" +rm -f $seq.full +status=0
tests/qemu-iotests/tests/iov-padding.out new file +59 -0
59 lines not on this page · raw message
Signature · 1 line
2.39.1