1. Patchew
  2. linux
  3. [PATCH v2 0/2] md: fix sync_action show
  4. View patch

[PATCH v2 2/2] md: fix sync_action incorrect display during resync

Zheng Qixing posted 2 patches 6 months ago
There is a newer version of this series
[PATCH v2 2/2] md: fix sync_action incorrect display during resync
Posted by Zheng Qixing 6 months ago 
From: Zheng Qixing <zhengqixing@huawei.com>

During raid resync, if a disk becomes faulty, the operation is
briefly interrupted. The MD_RECOVERY_RECOVER flag triggered by
the disk failure causes sync_action to incorrectly show "recover"
instead of "resync". The same issue affects reshape operations.

Reproduction steps:
  mdadm -Cv /dev/md1 -l1 -n4 -e1.2 /dev/sd{a..d} // -> resync happended
  mdadm -f /dev/md1 /dev/sda                     // -> resync interrupted
  cat sync_action
  -> recover

Add progress checks in md_sync_action() for resync/recover/reshape
to ensure the interface correctly reports the actual operation type.

Fixes: 4b10a3bc67c1 ("md: ensure resync is prioritized over recovery")
Signed-off-by: Zheng Qixing <zhengqixing@huawei.com>
---
 drivers/md/md.c | 38 ++++++++++++++++++++++++++++++++++++--
 1 file changed, 36 insertions(+), 2 deletions(-)

diff --git a/drivers/md/md.c b/drivers/md/md.c
index 4ea956a80343..798428d0870b 100644
--- a/drivers/md/md.c
+++ b/drivers/md/md.c
@@ -4845,9 +4845,34 @@ static bool rdev_needs_recovery(struct md_rdev *rdev, sector_t sectors)
 	return false;
 }
 
+static enum sync_action md_get_active_sync_action(struct mddev *mddev)
+{
+	struct md_rdev *rdev;
+	bool is_recover = false;
+
+	if (mddev->resync_offset < MaxSector)
+		return ACTION_RESYNC;
+
+	if (mddev->reshape_position != MaxSector)
+		return ACTION_RESHAPE;
+
+	rcu_read_lock();
+	rdev_for_each_rcu(rdev, mddev) {
+		if (rdev->raid_disk >= 0 &&
+		    rdev_needs_recovery(rdev, MaxSector)) {
+			is_recover = true;
+			break;
+		}
+	}
+	rcu_read_unlock();
+
+	return is_recover ? ACTION_RECOVER : ACTION_IDLE;
+}
+
 enum sync_action md_sync_action(struct mddev *mddev)
 {
 	unsigned long recovery = mddev->recovery;
+	enum sync_action active_action;
 
 	/*
 	 * frozen has the highest priority, means running sync_thread will be
@@ -4871,8 +4896,17 @@ enum sync_action md_sync_action(struct mddev *mddev)
 	    !test_bit(MD_RECOVERY_NEEDED, &recovery))
 		return ACTION_IDLE;
 
-	if (test_bit(MD_RECOVERY_RESHAPE, &recovery) ||
-	    mddev->reshape_position != MaxSector)
+	/*
+	 * Check if any sync operation (resync/recover/reshape) is
+	 * currently active. This ensures that only one sync operation
+	 * can run at a time. Returns the type of active operation, or
+	 * ACTION_IDLE if none are active.
+	 */
+	active_action = md_get_active_sync_action(mddev);
+	if (active_action != ACTION_IDLE)
+		return active_action;
+
+	if (test_bit(MD_RECOVERY_RESHAPE, &recovery))
 		return ACTION_RESHAPE;
 
 	if (test_bit(MD_RECOVERY_RECOVER, &recovery))
-- 
2.39.2
Re: [PATCH v2 2/2] md: fix sync_action incorrect display during resync
Posted by Paul Menzel 6 months ago 
Dear Zheng,


Thank you for your patch.

Am 12.08.25 um 04:17 schrieb Zheng Qixing:
> From: Zheng Qixing <zhengqixing@huawei.com>
> 
> During raid resync, if a disk becomes faulty, the operation is
> briefly interrupted. The MD_RECOVERY_RECOVER flag triggered by
> the disk failure causes sync_action to incorrectly show "recover"
> instead of "resync". The same issue affects reshape operations.
> 
> Reproduction steps:
>    mdadm -Cv /dev/md1 -l1 -n4 -e1.2 /dev/sd{a..d} // -> resync happended
>    mdadm -f /dev/md1 /dev/sda                     // -> resync interrupted
>    cat sync_action
>    -> recover
> 
> Add progress checks in md_sync_action() for resync/recover/reshape
> to ensure the interface correctly reports the actual operation type.
> 
> Fixes: 4b10a3bc67c1 ("md: ensure resync is prioritized over recovery")
> Signed-off-by: Zheng Qixing <zhengqixing@huawei.com>
> ---
>   drivers/md/md.c | 38 ++++++++++++++++++++++++++++++++++++--
>   1 file changed, 36 insertions(+), 2 deletions(-)
> 
> diff --git a/drivers/md/md.c b/drivers/md/md.c
> index 4ea956a80343..798428d0870b 100644
> --- a/drivers/md/md.c
> +++ b/drivers/md/md.c
> @@ -4845,9 +4845,34 @@ static bool rdev_needs_recovery(struct md_rdev *rdev, sector_t sectors)
>   	return false;
>   }
>   
> +static enum sync_action md_get_active_sync_action(struct mddev *mddev)
> +{
> +	struct md_rdev *rdev;
> +	bool is_recover = false;

`is_recover` sounds strange to me, but I am not an expert with the code. 
Maybe `needs_recovery`?

> +
> +	if (mddev->resync_offset < MaxSector)
> +		return ACTION_RESYNC;
> +
> +	if (mddev->reshape_position != MaxSector)
> +		return ACTION_RESHAPE;
> +
> +	rcu_read_lock();
> +	rdev_for_each_rcu(rdev, mddev) {
> +		if (rdev->raid_disk >= 0 &&
> +		    rdev_needs_recovery(rdev, MaxSector)) {
> +			is_recover = true;
> +			break;
> +		}
> +	}
> +	rcu_read_unlock();
> +
> +	return is_recover ? ACTION_RECOVER : ACTION_IDLE;
> +}
> +
>   enum sync_action md_sync_action(struct mddev *mddev)
>   {
>   	unsigned long recovery = mddev->recovery;
> +	enum sync_action active_action;
>   
>   	/*
>   	 * frozen has the highest priority, means running sync_thread will be
> @@ -4871,8 +4896,17 @@ enum sync_action md_sync_action(struct mddev *mddev)
>   	    !test_bit(MD_RECOVERY_NEEDED, &recovery))
>   		return ACTION_IDLE;
>   
> -	if (test_bit(MD_RECOVERY_RESHAPE, &recovery) ||
> -	    mddev->reshape_position != MaxSector)
> +	/*
> +	 * Check if any sync operation (resync/recover/reshape) is
> +	 * currently active. This ensures that only one sync operation
> +	 * can run at a time. Returns the type of active operation, or
> +	 * ACTION_IDLE if none are active.
> +	 */
> +	active_action = md_get_active_sync_action(mddev);
> +	if (active_action != ACTION_IDLE)
> +		return active_action;
> +
> +	if (test_bit(MD_RECOVERY_RESHAPE, &recovery))
>   		return ACTION_RESHAPE;
>   
>   	if (test_bit(MD_RECOVERY_RECOVER, &recovery))

Reviewed-by: Paul Menzel <pmenzel@molgen.mpg.de>


Kind regards,

Paul
Re: [PATCH v2 2/2] md: fix sync_action incorrect display during resync
Posted by Zheng Qixing 6 months ago 
Hi,


在 2025/8/12 17:22, Paul Menzel 写道:
> Dear Zheng,
>
>
> Thank you for your patch.
>
> Am 12.08.25 um 04:17 schrieb Zheng Qixing:
>> From: Zheng Qixing <zhengqixing@huawei.com>
>>
>> During raid resync, if a disk becomes faulty, the operation is
>> briefly interrupted. The MD_RECOVERY_RECOVER flag triggered by
>> the disk failure causes sync_action to incorrectly show "recover"
>> instead of "resync". The same issue affects reshape operations.
>>
>> Reproduction steps:
>>    mdadm -Cv /dev/md1 -l1 -n4 -e1.2 /dev/sd{a..d} // -> resync happended
>>    mdadm -f /dev/md1 /dev/sda                     // -> resync 
>> interrupted
>>    cat sync_action
>>    -> recover
>>
>> Add progress checks in md_sync_action() for resync/recover/reshape
>> to ensure the interface correctly reports the actual operation type.
>>
>> Fixes: 4b10a3bc67c1 ("md: ensure resync is prioritized over recovery")
>> Signed-off-by: Zheng Qixing <zhengqixing@huawei.com>
>> ---
>>   drivers/md/md.c | 38 ++++++++++++++++++++++++++++++++++++--
>>   1 file changed, 36 insertions(+), 2 deletions(-)
>>
>> diff --git a/drivers/md/md.c b/drivers/md/md.c
>> index 4ea956a80343..798428d0870b 100644
>> --- a/drivers/md/md.c
>> +++ b/drivers/md/md.c
>> @@ -4845,9 +4845,34 @@ static bool rdev_needs_recovery(struct md_rdev 
>> *rdev, sector_t sectors)
>>       return false;
>>   }
>>   +static enum sync_action md_get_active_sync_action(struct mddev 
>> *mddev)
>> +{
>> +    struct md_rdev *rdev;
>> +    bool is_recover = false;
>
> `is_recover` sounds strange to me, but I am not an expert with the 
> code. Maybe `needs_recovery`?


is_recover is used here to distinguish whether the current sync_action 
is a recover, rather than a resync or reshape.

But it's not a big deal, no need to focus on it :)


>
>> +
>> +    if (mddev->resync_offset < MaxSector)
>> +        return ACTION_RESYNC;
>> +
>> +    if (mddev->reshape_position != MaxSector)
>> +        return ACTION_RESHAPE;
>> +
>> +    rcu_read_lock();
>> +    rdev_for_each_rcu(rdev, mddev) {
>> +        if (rdev->raid_disk >= 0 &&
>> +            rdev_needs_recovery(rdev, MaxSector)) {
>> +            is_recover = true;
>> +            break;
>> +        }
>> +    }
>> +    rcu_read_unlock();
>> +
>> +    return is_recover ? ACTION_RECOVER : ACTION_IDLE;
>> +}
>> +
>>   enum sync_action md_sync_action(struct mddev *mddev)
>>   {
>>       unsigned long recovery = mddev->recovery;
>> +    enum sync_action active_action;
>>         /*
>>        * frozen has the highest priority, means running sync_thread 
>> will be
>> @@ -4871,8 +4896,17 @@ enum sync_action md_sync_action(struct mddev 
>> *mddev)
>>           !test_bit(MD_RECOVERY_NEEDED, &recovery))
>>           return ACTION_IDLE;
>>   -    if (test_bit(MD_RECOVERY_RESHAPE, &recovery) ||
>> -        mddev->reshape_position != MaxSector)
>> +    /*
>> +     * Check if any sync operation (resync/recover/reshape) is
>> +     * currently active. This ensures that only one sync operation
>> +     * can run at a time. Returns the type of active operation, or
>> +     * ACTION_IDLE if none are active.
>> +     */
>> +    active_action = md_get_active_sync_action(mddev);
>> +    if (active_action != ACTION_IDLE)
>> +        return active_action;
>> +
>> +    if (test_bit(MD_RECOVERY_RESHAPE, &recovery))
>>           return ACTION_RESHAPE;
>>         if (test_bit(MD_RECOVERY_RECOVER, &recovery))
>
> Reviewed-by: Paul Menzel <pmenzel@molgen.mpg.de>
>
>
> Kind regards,
>
> Paul


Thanks,

Qixing

Patchew 2.1-devel-b67e4c2

© 2016 - 2026 Red Hat, Inc.