block: add documentation for io_timeout
Subsystems (2)
PATCH v3 1/1 block: add documentation for io_timeout
Add documentation for /sys/block/<disk>/queue/io_timeout.
--- Documentation/ABI/testing/sysfs-block | 10 ++++++++++ Documentation/block/queue-sysfs.txt | 7 +++++++ 2 files changed, 17 insertions(+)
Documentation/ABI/testing/sysfs-block +10 -0
@@ -271,3 +271,13 @@ Description: size of 512B sectors of the zones of the device, with the eventual exception of the last zone of the device which may be smaller. + +What: /sys/block/<disk>/queue/io_timeout +Date: November 2018 +Contact: Weiping Zhang <[email protected]> +Description: + io_timeout is a request’s timeouts at block layer in + milliseconds. When the underlying driver starts processing + a request, the generic block layer will start a timer, if + this request cannot be completed in io_timeout milliseconds, + a timeout event will occur.
Documentation/block/queue-sysfs.txt +7 -0
@@ -67,6 +67,13 @@ If set to a value larger than 0, the kernel will put the process issuing IO to sleep for this amont of microseconds before entering classic polling. +io_timeout (RW) +--------------- +This is a request’s timeouts at block layer in milliseconds. When the +underlying driver starts processing a request, the generic block layer +will start a timer, if this request cannot be completed in io_timeout +milliseconds, a timeout event will occur. + iostats (RW) ------------- This file is used to control (on/off) the iostats accounting of the
Signature · 1 line
2.14.1
Can you please also send a patch to not show this attribute for drivers without a timeout handler? Thanks!
Quoted 47 lines · Weiping Zhang
Add documentation for /sys/block/<disk>/queue/io_timeout.
Signed-off-by Weiping Zhang <[email protected]>--- Documentation/ABI/testing/sysfs-block | 10 ++++++++++ Documentation/block/queue-sysfs.txt | 7 +++++++ 2 files changed, 17 insertions(+)Documentation/ABI/testing/sysfs-block +10
-0@@ -271,3 +271,13 @@ Description: size of 512B sectors of the zones of the device, with the eventual exception of the last zone of the device which may be smaller. + +What: /sys/block/<disk>/queue/io_timeout +Date: November 2018 +Contact: Weiping Zhang <[email protected]> +Description: + io_timeout is a request’s timeouts at block layer in + milliseconds. When the underlying driver starts processing + a request, the generic block layer will start a timer, if + this request cannot be completed in io_timeout milliseconds, + a timeout event will occur.Documentation/block/queue-sysfs.txt +7
-0@@ -67,6 +67,13 @@ If set to a value larger than 0, the kernel will put the process issuing IO to sleep for this amont of microseconds before entering classic polling. +io_timeout (RW) +--------------- +This is a request’s timeouts at block layer in milliseconds. When the +underlying driver starts processing a request, the generic block layer +will start a timer, if this request cannot be completed in io_timeout +milliseconds, a timeout event will occur. + iostats (RW) ------------- This file is used to control (on/off) the iostats accounting of theSignature · 1 line
2.14.1
---end quoted text---
Quoted 4 lines · Christoph Hellwig <[email protected]> 于2018年12月5日周三 下午10:40写道：
Can you please also send a patch to not show this attribute for drivers without a timeout handler? Thanks!
Do you means checking q->mq_ops->timeout is NULL or not ?
static void blk_mq_rq_timed_out(struct request *req, bool reserved)
{
req->rq_flags |= RQF_TIMED_OUT;If it's NULL, only re-start a timer for this request in blk_add_timer. Is there some defect ? if show this attribute and the driver doesn't support timeout handler ？
if (req->q->mq_ops->timeout) {
enum blk_eh_timer_return ret;
ret = req->q->mq_ops->timeout(req, reserved);
if (ret == BLK_EH_DONE)
return;
WARN_ON_ONCE(ret != BLK_EH_RESET_TIMER);
}
blk_add_timer(req);
}Quoted 49 lines · Weiping Zhang
Quoted 45 lines
Add documentation for /sys/block/<disk>/queue/io_timeout.
Signed-off-by Weiping Zhang <[email protected]>--- Documentation/ABI/testing/sysfs-block | 10 ++++++++++ Documentation/block/queue-sysfs.txt | 7 +++++++ 2 files changed, 17 insertions(+)Documentation/ABI/testing/sysfs-block +10
-0@@ -271,3 +271,13 @@ Description: size of 512B sectors of the zones of the device, with the eventual exception of the last zone of the device which may be smaller. + +What: /sys/block/<disk>/queue/io_timeout +Date: November 2018 +Contact: Weiping Zhang <[email protected]> +Description: + io_timeout is a request’s timeouts at block layer in + milliseconds. When the underlying driver starts processing + a request, the generic block layer will start a timer, if + this request cannot be completed in io_timeout milliseconds, + a timeout event will occur.Documentation/block/queue-sysfs.txt +7
-1@@ -67,6 +67,13 @@ If set to a value larger than 0, the kernel will put the process issuing IO to sleep for this amont of microseconds before entering classic polling. +io_timeout (RW) +--------------- +This is a request’s timeouts at block layer in milliseconds. When the +underlying driver starts processing a request, the generic block layer +will start a timer, if this request cannot be completed in io_timeout +milliseconds, a timeout event will occur. + iostats (RW) ------------- This file is used to control (on/off) the iostats accounting of the --Quoted 2 lines
2.14.1
---end quoted text---
Quoted 6 lines · Weiping Zhang <[email protected]> 于2018年12月5日周三 下午10:49写道：
Quoted 4 lines · Christoph Hellwig <[email protected]> 于2018年12月5日周三 下午10:40写道：
Can you please also send a patch to not show this attribute for drivers without a timeout handler? Thanks!
Is there a simple way do that ?
Shall we return -ENOTSUPP when user read/write this attribute when driver has no timeout handler ?
Quoted 71 lines
Do you means checking
q->mq_ops->timeoutis NULL or not ?static void
blk_mq_rq_timed_out(struct request *req, bool reserved){ req->rq_flags |= RQF_TIMED_OUT;If it's NULL, only re-start a timer for this request in
blk_add_timer. Is there some defect ? if show this attribute and the driver doesn't support timeout handler ？if (req->q->mq_ops->timeout) { enum blk_eh_timer_return ret; ret = req->q->mq_ops->timeout(req, reserved); if (ret == BLK_EH_DONE) return; WARN_ON_ONCE(ret != BLK_EH_RESET_TIMER); } blk_add_timer(req); }Quoted 49 lines · Weiping Zhang
Quoted 45 lines
Add documentation for /sys/block/<disk>/queue/io_timeout.
Signed-off-by Weiping Zhang <[email protected]>--- Documentation/ABI/testing/sysfs-block | 10 ++++++++++ Documentation/block/queue-sysfs.txt | 7 +++++++ 2 files changed, 17 insertions(+)Documentation/ABI/testing/sysfs-block +10
-0@@ -271,3 +271,13 @@ Description: size of 512B sectors of the zones of the device, with the eventual exception of the last zone of the device which may be smaller. + +What: /sys/block/<disk>/queue/io_timeout +Date: November 2018 +Contact: Weiping Zhang <[email protected]> +Description: + io_timeout is a request’s timeouts at block layer in + milliseconds. When the underlying driver starts processing + a request, the generic block layer will start a timer, if + this request cannot be completed in io_timeout milliseconds, + a timeout event will occur.Documentation/block/queue-sysfs.txt +7
-1@@ -67,6 +67,13 @@ If set to a value larger than 0, the kernel will put the process issuing IO to sleep for this amont of microseconds before entering classic polling. +io_timeout (RW) +--------------- +This is a request’s timeouts at block layer in milliseconds. When the +underlying driver starts processing a request, the generic block layer +will start a timer, if this request cannot be completed in io_timeout +milliseconds, a timeout event will occur. + iostats (RW) ------------- This file is used to control (on/off) the iostats accounting of the --Quoted 2 lines
2.14.1
---end quoted text---
Quoted 3 lines · Weiping Zhang
Weiping Zhang <[email protected]>于2018年12月5日周三下午10:49写道：>>Christoph Hellwig <[email protected]>于2018年12月5日周三下午10:40写道：>>>Can you please also send a patch to not show this attribute for
Quoted 1 line
Quoted 1 line
drivers without a timeout handler? Thanks!>
Is there a simple way do that ?
How about checking the timeout member of struct blk_mq_ops for blk-mq and checking the rq_timed_out_fn member in struct request_queue for the legacy block layer?
Quoted 2 lines
Shall we return
-ENOTSUPPwhen user read/write this attribute when driver has no timeout handler ?
A much more elegant solution is to introduce a sysfs attribute group for the io_timeout attribute and to make that group visible only if a timeout handler has been defined. See e.g. disk_attr_group in block/genhd.c for an example.
Bart.
Quoted 11 lines · Bart Van Assche
Quoted 6 lines · Weiping Zhang
Quoted 3 lines · Weiping Zhang <[email protected]> 于2018年12月5日周三 下午10:49写道：
Quoted 2 lines · Christoph Hellwig <[email protected]> 于2018年12月5日周三 下午10:40写道：
Can you please also send a patch to not show this attribute for drivers without a timeout handler? Thanks!
Is there a simple way do that ?
How about checking the timeout member of struct
blk_mq_opsfor blk-mq and checking therq_timed_out_fnmember in structrequest_queuefor the legacy block layer?
Just the former given that the legacy code is gone in for-next.
Quoted 6 lines
Quoted 2 lines
Shall we return
-ENOTSUPPwhen user read/write this attribute when driver has no timeout handler ?A much more elegant solution is to introduce a sysfs attribute group for the
io_timeoutattribute and to make that group visible only if a timeout handler has been defined. See e.g.disk_attr_groupinblock/genhd.cfor an example.
Agreed, that is the way to go.
Quoted 6 lines · Weiping Zhang
+-Description: +- io_timeout is a request’s timeouts at block layer in +- milliseconds. When the underlying driver starts processing +- a request, the generic block layer will start a timer, if +- this request cannot be completed in io_timeout milliseconds, +- a timeout event will occur.
Sorry but I think this description is still somewhat confusing. How about changing that description into the following?
io_timeout is the request timeout in milliseconds. If a request does not complete in this time then the block driver timeout handler is invoked. That timeout handler can decide to retry the request, to fail it or to start a device recovery strategy.
Bart.
Quoted 9 lines · Bart Van Assche <[email protected]> 于2018年12月7日周五 上午12:22写道：
Documentation/ABI/testing/sysfs-block Description: · line 283
Quoted 6 lines · Weiping Zhang
Documentation/ABI/testing/sysfs-block +6
-0+Description: + io_timeout is a request’s timeouts at block layer in + milliseconds. When the underlying driver starts processing + a request, the generic block layer will start a timer, if + this request cannot be completed in io_timeout milliseconds, + a timeout event will occur.
Quoted 8 lines
Sorry but I think this description is still somewhat confusing. How about changing that description into the following?
io_timeoutis the request timeout in milliseconds. If a request does not complete in this time then the block driver timeout handler is invoked. That timeout handler can decide to retry the request, to fail it or to start a device recovery strategy.
Sorry for late reply, thanks for your suggestion I'll post it in V4.
Quoted 10 lines
Bart.
Quoted 5 lines
Quoted 1 line
Is there a simple way do that ?
How about checking the timeout member of struct
blk_mq_opsfor blk-mq and checking therq_timed_out_fnmember in structrequest_queuefor the legacy block layer?Just the former given that the legacy code is gone in for-next.
Quoted 8 lines
Quoted 6 lines
Quoted 2 lines
Shall we return
-ENOTSUPPwhen user read/write this attribute when driver has no timeout handler ?A much more elegant solution is to introduce a sysfs attribute group for the
io_timeoutattribute and to make that group visible only if a timeout handler has been defined. See e.g.disk_attr_groupinblock/genhd.cfor an example.Agreed, that is the way to go.
OK, I'll do that.
Thanks
Weiping