- Linux-stable-mirror - lists.linaro.org

[Patch] sched/fair: Avoid throttle_list starvation with low cfs quota

by Phil Auld

From: "Phil Auld" <pauld(a)redhat.com> sched/fair: Avoid throttle_list starvation with low cfs quota With a very low cpu.cfs_quota_us setting, such as the minimum of 1000, distribute_cfs_runtime may not empty the throttled_list before it runs out of runtime to distribute. In that case, due to the change from c06f04c7048 to put throttled entries at the head of the list, later entries on the list will starve. Essentially, the same X processes will get pulled off the list, given CPU time and then, when expired, get put back on the head of the list where distribute_cfs_runtime will give runtime to the same set of processes leaving the rest. Fix the issue by setting a bit in struct cfs_bandwidth when distribute_cfs_runtime is running, so that the code in throttle_cfs_rq can decide to put the throttled entry on the tail or the head of the list. The bit is set/cleared by the callers of distribute_cfs_runtime while they hold cfs_bandwidth->lock. Signed-off-by: Phil Auld <pauld(a)redhat.com> Fixes: c06f04c70489 ("sched: Fix potential near-infinite distribute_cfs_runtime() loop") Cc: Peter Zijlstra <peterz(a)infradead.org> Cc: Ingo Molnar <mingo(a)kernel.org> Cc: stable(a)vger.kernel.org --- This is easy to reproduce with a handful of cpu consumers. I use crash on the live system. In some cases you can simply look at the throttled list and see the later entries are not changing: crash> list cfs_rq.throttled_list -H 0xffff90b54f6ade40 -s cfs_rq.runtime_remaining | paste - - | awk '{print $1" "$4}' | pr -t -n3 1 ffff90b56cb2d200 -976050 2 ffff90b56cb2cc00 -484925 3 ffff90b56cb2bc00 -658814 4 ffff90b56cb2ba00 -275365 5 ffff90b166a45600 -135138 6 ffff90b56cb2da00 -282505 7 ffff90b56cb2e000 -148065 8 ffff90b56cb2fa00 -872591 9 ffff90b56cb2c000 -84687 10 ffff90b56cb2f000 -87237 11 ffff90b166a40a00 -164582 crash> list cfs_rq.throttled_list -H 0xffff90b54f6ade40 -s cfs_rq.runtime_remaining | paste - - | awk '{print $1" "$4}' | pr -t -n3 1 ffff90b56cb2d200 -994147 2 ffff90b56cb2cc00 -306051 3 ffff90b56cb2bc00 -961321 4 ffff90b56cb2ba00 -24490 5 ffff90b166a45600 -135138 6 ffff90b56cb2da00 -282505 7 ffff90b56cb2e000 -148065 8 ffff90b56cb2fa00 -872591 9 ffff90b56cb2c000 -84687 10 ffff90b56cb2f000 -87237 11 ffff90b166a40a00 -164582 Sometimes it is easier to see by finding a process getting starved and looking at the sched_info: crash> task ffff8eb765994500 sched_info PID: 7800 TASK: ffff8eb765994500 CPU: 16 COMMAND: "cputest" sched_info = { pcount = 8, run_delay = 697094208, last_arrival = 240260125039, last_queued = 240260327513 }, crash> task ffff8eb765994500 sched_info PID: 7800 TASK: ffff8eb765994500 CPU: 16 COMMAND: "cputest" sched_info = { pcount = 8, run_delay = 697094208, last_arrival = 240260125039, last_queued = 240260327513 }, fair.c | 22 +++++++++++++++++++--- sched.h | 2 ++ 2 files changed, 21 insertions(+), 3 deletions(-) diff --git a/kernel/sched/fair.c b/kernel/sched/fair.c index 7fc4a371bdd2..f88e00705b55 100644 --- a/kernel/sched/fair.c +++ b/kernel/sched/fair.c @@ -4476,9 +4476,13 @@ static void throttle_cfs_rq(struct cfs_rq *cfs_rq) /* * Add to the _head_ of the list, so that an already-started - * distribute_cfs_runtime will not see us + * distribute_cfs_runtime will not see us. If disribute_cfs_runtime is + * not running add to the tail so that later runqueues don't get starved. */ - list_add_rcu(&cfs_rq->throttled_list, &cfs_b->throttled_cfs_rq); + if (cfs_b->distribute_running) + list_add_rcu(&cfs_rq->throttled_list, &cfs_b->throttled_cfs_rq); + else + list_add_tail_rcu(&cfs_rq->throttled_list, &cfs_b->throttled_cfs_rq); /* * If we're the first throttled task, make sure the bandwidth @@ -4622,14 +4626,16 @@ static int do_sched_cfs_period_timer(struct cfs_bandwidth *cfs_b, int overrun) * in us over-using our runtime if it is all used during this loop, but * only by limited amounts in that extreme case. */ - while (throttled && cfs_b->runtime > 0) { + while (throttled && cfs_b->runtime > 0 && !cfs_b->distribute_running) { runtime = cfs_b->runtime; + cfs_b->distribute_running = 1; raw_spin_unlock(&cfs_b->lock); /* we can't nest cfs_b->lock while distributing bandwidth */ runtime = distribute_cfs_runtime(cfs_b, runtime, runtime_expires); raw_spin_lock(&cfs_b->lock); + cfs_b->distribute_running = 0; throttled = !list_empty(&cfs_b->throttled_cfs_rq); cfs_b->runtime -= min(runtime, cfs_b->runtime); @@ -4740,6 +4746,11 @@ static void do_sched_cfs_slack_timer(struct cfs_bandwidth *cfs_b) /* confirm we're still not at a refresh boundary */ raw_spin_lock(&cfs_b->lock); + if (cfs_b->distribute_running) { + raw_spin_unlock(&cfs_b->lock); + return; + } + if (runtime_refresh_within(cfs_b, min_bandwidth_expiration)) { raw_spin_unlock(&cfs_b->lock); return; @@ -4749,6 +4760,9 @@ static void do_sched_cfs_slack_timer(struct cfs_bandwidth *cfs_b) runtime = cfs_b->runtime; expires = cfs_b->runtime_expires; + if (runtime) + cfs_b->distribute_running = 1; + raw_spin_unlock(&cfs_b->lock); if (!runtime) @@ -4759,6 +4773,7 @@ static void do_sched_cfs_slack_timer(struct cfs_bandwidth *cfs_b) raw_spin_lock(&cfs_b->lock); if (expires == cfs_b->runtime_expires) cfs_b->runtime -= min(runtime, cfs_b->runtime); + cfs_b->distribute_running = 0; raw_spin_unlock(&cfs_b->lock); } @@ -4867,6 +4882,7 @@ void init_cfs_bandwidth(struct cfs_bandwidth *cfs_b) cfs_b->period_timer.function = sched_cfs_period_timer; hrtimer_init(&cfs_b->slack_timer, CLOCK_MONOTONIC, HRTIMER_MODE_REL); cfs_b->slack_timer.function = sched_cfs_slack_timer; + cfs_b->distribute_running = 0; } static void init_cfs_rq_runtime(struct cfs_rq *cfs_rq) diff --git a/kernel/sched/sched.h b/kernel/sched/sched.h index 455fa330de04..9683f458aec7 100644 --- a/kernel/sched/sched.h +++ b/kernel/sched/sched.h @@ -346,6 +346,8 @@ struct cfs_bandwidth { int nr_periods; int nr_throttled; u64 throttled_time; + + bool distribute_running; #endif }; --

7 years

4
5
0 0

v4.14.80 build: 0 failures 0 warnings (v4.14.80)

by Build bot for Mark Brown

Tree/Branch: v4.14.80 Git describe: v4.14.80 Commit: 0b047cbc44 Linux 4.14.80 Build Time: 112 min 37 sec Passed: 11 / 11 (100.00 %) Failed: 0 / 11 ( 0.00 %) Errors: 0 Warnings: 0 Section Mismatches: 0 ------------------------------------------------------------------------------- defconfigs with issues (other than build errors): ------------------------------------------------------------------------------- =============================================================================== Detailed per-defconfig build reports below: ------------------------------------------------------------------------------- Passed with no errors, warnings or mismatches: arm64-allnoconfig arm64-allmodconfig arm-multi_v5_defconfig arm-multi_v7_defconfig x86_64-defconfig arm-allmodconfig arm-allnoconfig x86_64-allnoconfig arm-multi_v4t_defconfig x86_64-allmodconfig arm64-defconfig close failed in file object destructor: sys.excepthook is missing lost sys.stderr

7 years

1
0
0 0

[4.4.y v3] ext4: avoid running out of journal credits when appending to an inline file

by Chenbo Feng

From: Theodore Ts'o <tytso(a)mit.edu> commit 8bc1379b82b8e809eef77a9fedbb75c6c297be19 upstream. Use a separate journal transaction if it turns out that we need to convert an inline file to use an data block. Otherwise we could end up failing due to not having journal credits. This addresses CVE-2018-10883. https://bugzilla.kernel.org/show_bug.cgi?id=200071 Change-Id: Ifbe92e379f7a25fb252a2584356ccb91f902ea8f Signed-off-by: Theodore Ts'o <tytso(a)mit.edu> Cc: stable(a)kernel.org [fengc(a)google.com: 4.4 and 4.9 backport: adjust context] Signed-off-by: Chenbo Feng <fengc(a)google.com> --- fs/ext4/ext4.h | 3 --- fs/ext4/inline.c | 38 +------------------------------------- fs/ext4/xattr.c | 18 ++---------------- 3 files changed, 3 insertions(+), 56 deletions(-) diff --git a/fs/ext4/ext4.h b/fs/ext4/ext4.h index f5d9f82b173a..b6e25d771eea 100644 --- a/fs/ext4/ext4.h +++ b/fs/ext4/ext4.h @@ -3039,9 +3039,6 @@ extern struct buffer_head *ext4_get_first_inline_block(struct inode *inode, extern int ext4_inline_data_fiemap(struct inode *inode, struct fiemap_extent_info *fieinfo, int *has_inline, __u64 start, __u64 len); -extern int ext4_try_to_evict_inline_data(handle_t *handle, - struct inode *inode, - int needed); extern void ext4_inline_data_truncate(struct inode *inode, int *has_inline); extern int ext4_convert_inline_data(struct inode *inode); diff --git a/fs/ext4/inline.c b/fs/ext4/inline.c index 1e7a9774119c..5ead3b0f3d34 100644 --- a/fs/ext4/inline.c +++ b/fs/ext4/inline.c @@ -888,11 +888,11 @@ retry_journal: flags |= AOP_FLAG_NOFS; if (ret == -ENOSPC) { + ext4_journal_stop(handle); ret = ext4_da_convert_inline_data_to_extent(mapping, inode, flags, fsdata); - ext4_journal_stop(handle); if (ret == -ENOSPC && ext4_should_retry_alloc(inode->i_sb, &retries)) goto retry_journal; @@ -1867,42 +1867,6 @@ out: return (error < 0 ? error : 0); } -/* - * Called during xattr set, and if we can sparse space 'needed', - * just create the extent tree evict the data to the outer block. - * - * We use jbd2 instead of page cache to move data to the 1st block - * so that the whole transaction can be committed as a whole and - * the data isn't lost because of the delayed page cache write. - */ -int ext4_try_to_evict_inline_data(handle_t *handle, - struct inode *inode, - int needed) -{ - int error; - struct ext4_xattr_entry *entry; - struct ext4_inode *raw_inode; - struct ext4_iloc iloc; - - error = ext4_get_inode_loc(inode, &iloc); - if (error) - return error; - - raw_inode = ext4_raw_inode(&iloc); - entry = (struct ext4_xattr_entry *)((void *)raw_inode + - EXT4_I(inode)->i_inline_off); - if (EXT4_XATTR_LEN(entry->e_name_len) + - EXT4_XATTR_SIZE(le32_to_cpu(entry->e_value_size)) < needed) { - error = -ENOSPC; - goto out; - } - - error = ext4_convert_inline_data_nolock(handle, inode, &iloc); -out: - brelse(iloc.bh); - return error; -} - void ext4_inline_data_truncate(struct inode *inode, int *has_inline) { handle_t *handle; diff --git a/fs/ext4/xattr.c b/fs/ext4/xattr.c index d0aaf338fa9f..0d4b6743dc2f 100644 --- a/fs/ext4/xattr.c +++ b/fs/ext4/xattr.c @@ -1038,22 +1038,8 @@ int ext4_xattr_ibody_inline_set(handle_t *handle, struct inode *inode, if (EXT4_I(inode)->i_extra_isize == 0) return -ENOSPC; error = ext4_xattr_set_entry(i, s); - if (error) { - if (error == -ENOSPC && - ext4_has_inline_data(inode)) { - error = ext4_try_to_evict_inline_data(handle, inode, - EXT4_XATTR_LEN(strlen(i->name) + - EXT4_XATTR_SIZE(i->value_len))); - if (error) - return error; - error = ext4_xattr_ibody_find(inode, i, is); - if (error) - return error; - error = ext4_xattr_set_entry(i, s); - } - if (error) - return error; - } + if (error) + return error; header = IHDR(inode, ext4_raw_inode(&is->iloc)); if (!IS_LAST_ENTRY(s->first)) { header->h_magic = cpu_to_le32(EXT4_XATTR_MAGIC); -- 2.19.1.568.g152ad8e336-goog

7 years

2
2
0 0

Can I Trust You?

by Abel Brent

Dear friend, I am Abel Brent, a NATO soldier serving in Afghanistan. I and my Comrades, we are seeking your assistance to help us receive/invest our funds in your country in any lucrative business. Please if this proposal is acceptable by you, kindly respond back to me for more details. Thanks and waiting to hear from you Abel.

7 years

1
0
0 0

FAILED: patch "[PATCH] remoteproc: qcom: q6v5-mss: add SCM probe dependency" failed to apply to 4.18-stable tree

by gregkh＠linuxfoundation.org

The patch below does not apply to the 4.18-stable tree. If someone wants it applied there, or to any other stable or longterm tree, then please email the backport, including the original git commit id to <stable(a)vger.kernel.org>. thanks, greg k-h ------------------ original commit in Linus's tree ------------------ >From bbcda30271752bb7490f2e2aef5411dbcae69116 Mon Sep 17 00:00:00 2001 From: Brian Norris <briannorris(a)chromium.org> Date: Mon, 8 Oct 2018 19:08:05 -0700 Subject: [PATCH] remoteproc: qcom: q6v5-mss: add SCM probe dependency The memory ownership transfer request is performed using SCM, ensure that SCM is available before we probe the driver if memory protection is needed by the subsystem. Fixes: 6c5a9dc2481b ("remoteproc: qcom: Make secure world call for mem ownership switch") Cc: stable(a)vger.kernel.org Signed-off-by: Brian Norris <briannorris(a)chromium.org> [bjorn: Added condition for need_mem_protection, updated commit message] Signed-off-by: Bjorn Andersson <bjorn.andersson(a)linaro.org> diff --git a/drivers/remoteproc/qcom_q6v5_mss.c b/drivers/remoteproc/qcom_q6v5_mss.c index 3448f1df2e87..4c47f5e0a87c 100644 --- a/drivers/remoteproc/qcom_q6v5_mss.c +++ b/drivers/remoteproc/qcom_q6v5_mss.c @@ -1152,6 +1152,9 @@ static int q6v5_probe(struct platform_device *pdev) if (!desc) return -EINVAL; + if (desc->need_mem_protection && !qcom_scm_is_available()) + return -EPROBE_DEFER; + rproc = rproc_alloc(&pdev->dev, pdev->name, &q6v5_ops, desc->hexagon_mba_image, sizeof(*qproc)); if (!rproc) {

7 years

1
0
0 0

FAILED: patch "[PATCH] remoteproc: qcom: q6v5-mss: add SCM probe dependency" failed to apply to 4.19-stable tree

by gregkh＠linuxfoundation.org

The patch below does not apply to the 4.19-stable tree. If someone wants it applied there, or to any other stable or longterm tree, then please email the backport, including the original git commit id to <stable(a)vger.kernel.org>. thanks, greg k-h ------------------ original commit in Linus's tree ------------------ >From bbcda30271752bb7490f2e2aef5411dbcae69116 Mon Sep 17 00:00:00 2001 From: Brian Norris <briannorris(a)chromium.org> Date: Mon, 8 Oct 2018 19:08:05 -0700 Subject: [PATCH] remoteproc: qcom: q6v5-mss: add SCM probe dependency The memory ownership transfer request is performed using SCM, ensure that SCM is available before we probe the driver if memory protection is needed by the subsystem. Fixes: 6c5a9dc2481b ("remoteproc: qcom: Make secure world call for mem ownership switch") Cc: stable(a)vger.kernel.org Signed-off-by: Brian Norris <briannorris(a)chromium.org> [bjorn: Added condition for need_mem_protection, updated commit message] Signed-off-by: Bjorn Andersson <bjorn.andersson(a)linaro.org> diff --git a/drivers/remoteproc/qcom_q6v5_mss.c b/drivers/remoteproc/qcom_q6v5_mss.c index 3448f1df2e87..4c47f5e0a87c 100644 --- a/drivers/remoteproc/qcom_q6v5_mss.c +++ b/drivers/remoteproc/qcom_q6v5_mss.c @@ -1152,6 +1152,9 @@ static int q6v5_probe(struct platform_device *pdev) if (!desc) return -EINVAL; + if (desc->need_mem_protection && !qcom_scm_is_available()) + return -EPROBE_DEFER; + rproc = rproc_alloc(&pdev->dev, pdev->name, &q6v5_ops, desc->hexagon_mba_image, sizeof(*qproc)); if (!rproc) {

7 years

1
0
0 0

FAILED: patch "[PATCH] media: cec: forgot to cancel delayed work" failed to apply to 4.18-stable tree

by gregkh＠linuxfoundation.org

The patch below does not apply to the 4.18-stable tree. If someone wants it applied there, or to any other stable or longterm tree, then please email the backport, including the original git commit id to <stable(a)vger.kernel.org>. thanks, greg k-h ------------------ original commit in Linus's tree ------------------ >From 490d84f6d73c12f4204241cff8651eed60aae914 Mon Sep 17 00:00:00 2001 From: Hans Verkuil <hverkuil(a)xs4all.nl> Date: Mon, 15 Oct 2018 06:14:22 -0400 Subject: [PATCH] media: cec: forgot to cancel delayed work If the wait for completion was interrupted, then make sure to cancel any delayed work. This can only happen if a transmit is waiting for a reply, and you press Ctrl-C or reboot/poweroff or something like that which interrupts the thread waiting for the reply and then proceeds to delete the CEC message. Since the delayed work wasn't canceled, once it would trigger it referred to stale data and resulted in a kernel oops. Fixes: 7ec2b3b941a6 ("cec: add new tx/rx status bits to detect aborts/timeouts") Signed-off-by: Hans Verkuil <hans.verkuil(a)cisco.com> Cc: <stable(a)vger.kernel.org> # for v4.18 and up Signed-off-by: Mauro Carvalho Chehab <mchehab+samsung(a)kernel.org> diff --git a/drivers/media/cec/cec-adap.c b/drivers/media/cec/cec-adap.c index 0c0d9107383e..31d1f4ab915e 100644 --- a/drivers/media/cec/cec-adap.c +++ b/drivers/media/cec/cec-adap.c @@ -844,6 +844,8 @@ int cec_transmit_msg_fh(struct cec_adapter *adap, struct cec_msg *msg, */ mutex_unlock(&adap->lock); wait_for_completion_killable(&data->c); + if (!data->completed) + cancel_delayed_work_sync(&data->work); mutex_lock(&adap->lock); /* Cancel the transmit if it was interrupted */

7 years

1
0
0 0

FAILED: patch "[PATCH] media: cec: forgot to cancel delayed work" failed to apply to 4.19-stable tree

by gregkh＠linuxfoundation.org

The patch below does not apply to the 4.19-stable tree. If someone wants it applied there, or to any other stable or longterm tree, then please email the backport, including the original git commit id to <stable(a)vger.kernel.org>. thanks, greg k-h ------------------ original commit in Linus's tree ------------------ >From 490d84f6d73c12f4204241cff8651eed60aae914 Mon Sep 17 00:00:00 2001 From: Hans Verkuil <hverkuil(a)xs4all.nl> Date: Mon, 15 Oct 2018 06:14:22 -0400 Subject: [PATCH] media: cec: forgot to cancel delayed work If the wait for completion was interrupted, then make sure to cancel any delayed work. This can only happen if a transmit is waiting for a reply, and you press Ctrl-C or reboot/poweroff or something like that which interrupts the thread waiting for the reply and then proceeds to delete the CEC message. Since the delayed work wasn't canceled, once it would trigger it referred to stale data and resulted in a kernel oops. Fixes: 7ec2b3b941a6 ("cec: add new tx/rx status bits to detect aborts/timeouts") Signed-off-by: Hans Verkuil <hans.verkuil(a)cisco.com> Cc: <stable(a)vger.kernel.org> # for v4.18 and up Signed-off-by: Mauro Carvalho Chehab <mchehab+samsung(a)kernel.org> diff --git a/drivers/media/cec/cec-adap.c b/drivers/media/cec/cec-adap.c index 0c0d9107383e..31d1f4ab915e 100644 --- a/drivers/media/cec/cec-adap.c +++ b/drivers/media/cec/cec-adap.c @@ -844,6 +844,8 @@ int cec_transmit_msg_fh(struct cec_adapter *adap, struct cec_msg *msg, */ mutex_unlock(&adap->lock); wait_for_completion_killable(&data->c); + if (!data->completed) + cancel_delayed_work_sync(&data->work); mutex_lock(&adap->lock); /* Cancel the transmit if it was interrupted */

7 years

1
0
0 0

FAILED: patch "[PATCH] media: v4l2-tpg: fix kernel oops when enabling HFLIP and OSD" failed to apply to 4.4-stable tree

by gregkh＠linuxfoundation.org

The patch below does not apply to the 4.4-stable tree. If someone wants it applied there, or to any other stable or longterm tree, then please email the backport, including the original git commit id to <stable(a)vger.kernel.org>. thanks, greg k-h ------------------ original commit in Linus's tree ------------------ >From 250854eed5d45a73d81e4137dfd85180af6f2ec3 Mon Sep 17 00:00:00 2001 From: Hans Verkuil <hverkuil(a)xs4all.nl> Date: Mon, 8 Oct 2018 15:08:27 -0400 Subject: [PATCH] media: v4l2-tpg: fix kernel oops when enabling HFLIP and OSD When the OSD is on (i.e. vivid displays text on top of the test pattern), and you enable hflip, then the driver crashes. The cause turned out to be a division of a negative number by an unsigned value. You expect that -8 / 2U would be -4, but in reality it is 2147483644 :-( Fixes: 3e14e7a82c1ef ("vivid-tpg: add hor/vert downsampling support to tpg_gen_text") Signed-off-by: Hans Verkuil <hans.verkuil(a)cisco.com> Reported-by: Mauro Carvalho Chehab <mchehab+samsung(a)kernel.org> Cc: <stable(a)vger.kernel.org> # for v4.1 and up Signed-off-by: Mauro Carvalho Chehab <mchehab+samsung(a)kernel.org> diff --git a/drivers/media/common/v4l2-tpg/v4l2-tpg-core.c b/drivers/media/common/v4l2-tpg/v4l2-tpg-core.c index 76b125ebee6d..fa483b95bc5a 100644 --- a/drivers/media/common/v4l2-tpg/v4l2-tpg-core.c +++ b/drivers/media/common/v4l2-tpg/v4l2-tpg-core.c @@ -1801,7 +1801,7 @@ typedef struct { u16 __; u8 _; } __packed x24; pos[7] = (chr & (0x01 << 0) ? fg : bg); \ } \ \ - pos += (tpg->hflip ? -8 : 8) / hdiv; \ + pos += (tpg->hflip ? -8 : 8) / (int)hdiv; \ } \ } \ } while (0)

7 years

1
0
0 0

FAILED: patch "[PATCH] xprtrdma: Reset credit grant properly after a disconnect" failed to apply to 4.4-stable tree

by gregkh＠linuxfoundation.org

The patch below does not apply to the 4.4-stable tree. If someone wants it applied there, or to any other stable or longterm tree, then please email the backport, including the original git commit id to <stable(a)vger.kernel.org>. thanks, greg k-h ------------------ original commit in Linus's tree ------------------ >From ef739b2175dde9c05594f768cb78149f1ce2ac36 Mon Sep 17 00:00:00 2001 From: Chuck Lever <chuck.lever(a)oracle.com> Date: Mon, 1 Oct 2018 14:25:14 -0400 Subject: [PATCH] xprtrdma: Reset credit grant properly after a disconnect On a fresh connection, an RPC/RDMA client is supposed to send only one RPC Call until it gets a credit grant in the first RPC Reply from the server [RFC 8166, Section 3.3.3]. There is a bug in the Linux client's credit accounting mechanism introduced by commit e7ce710a8802 ("xprtrdma: Avoid deadlock when credit window is reset"). On connect, it simply dumps all pending RPC Calls onto the new connection. Servers have been tolerant of this bad behavior. Currently no server implementation ever changes its credit grant over reconnects, and servers always repost enough Receives before connections are fully established. To correct this issue, ensure that the client resets both the credit grant _and_ the congestion window when handling a reconnect. Fixes: e7ce710a8802 ("xprtrdma: Avoid deadlock when credit ... ") Signed-off-by: Chuck Lever <chuck.lever(a)oracle.com> Cc: stable(a)kernel.org Signed-off-by: Anna Schumaker <Anna.Schumaker(a)Netapp.com> diff --git a/net/sunrpc/xprtrdma/svc_rdma_backchannel.c b/net/sunrpc/xprtrdma/svc_rdma_backchannel.c index a68180090554..b9827665ff35 100644 --- a/net/sunrpc/xprtrdma/svc_rdma_backchannel.c +++ b/net/sunrpc/xprtrdma/svc_rdma_backchannel.c @@ -248,6 +248,7 @@ static void xprt_rdma_bc_close(struct rpc_xprt *xprt) { dprintk("svcrdma: %s: xprt %p\n", __func__, xprt); + xprt->cwnd = RPC_CWNDSHIFT; } static void diff --git a/net/sunrpc/xprtrdma/transport.c b/net/sunrpc/xprtrdma/transport.c index 143ce2579ba9..98cbc7b060ba 100644 --- a/net/sunrpc/xprtrdma/transport.c +++ b/net/sunrpc/xprtrdma/transport.c @@ -468,6 +468,12 @@ xprt_rdma_close(struct rpc_xprt *xprt) xprt->reestablish_timeout = 0; xprt_disconnect_done(xprt); rpcrdma_ep_disconnect(ep, ia); + + /* Prepare @xprt for the next connection by reinitializing + * its credit grant to one (see RFC 8166, Section 3.3.3). + */ + r_xprt->rx_buf.rb_credits = 1; + xprt->cwnd = RPC_CWNDSHIFT; } /**

7 years

1
0
0 0

2025

2024

2023

2022

2021

2020

2019

2018

2017

Linux-stable-mirror