[Ocfs2-devel] [PATCH] ocfs2: the ip_alloc_sem should be taken in ocfs2_get_block()

[Ocfs2-devel] [PATCH] ocfs2: the ip_alloc_sem should be taken in ocfs2_get_block()

[alex chen]

The ip_alloc_sem should be taken in ocfs2_get_block() when reading file
in DIRECT mode to prevent concurrent access to extent tree with
ocfs2_dio_end_io_write(), which may cause BUGON in
ocfs2_get_clusters_nocache()->BUG_ON(v_cluster < le32_to_cpu(rec->e_cpos))

Signed-off-by: Alex Chen <alex.chen@huawei.com>
Reviewed-by: Jun Piao <piaojun@huawei.com>

---
 fs/ocfs2/aops.c | 21 +++++++++++++++------
 1 file changed, 15 insertions(+), 6 deletions(-)

diff --git a/fs/ocfs2/aops.c b/fs/ocfs2/aops.c
index 88a31e9..5cb939f 100644
--- a/fs/ocfs2/aops.c
+++ b/fs/ocfs2/aops.c
@@ -134,6 +134,19 @@ static int ocfs2_symlink_get_block(struct inode *inode, sector_t iblock,
 	return err;
 }

+static int ocfs2_get_block_lock(struct inode *inode, sector_t iblock,
+		    struct buffer_head *bh_result, int create)
+{
+	int ret;
+	struct ocfs2_inode_info *oi = OCFS2_I(inode);
+
+	down_read(&oi->ip_alloc_sem);
+	ret = ocfs2_get_block(inode, iblock, bh_result, create);
+	up_read(&oi->ip_alloc_sem);
+
+	return ret;
+}
+
 int ocfs2_get_block(struct inode *inode, sector_t iblock,
 		    struct buffer_head *bh_result, int create)
 {
@@ -2154,12 +2167,8 @@ static int ocfs2_dio_get_block(struct inode *inode, sector_t iblock,
 	 * while file size will be changed.
 	 */
 	if (pos + total_len <= i_size_read(inode)) {
-		down_read(&oi->ip_alloc_sem);
 		/* This is the fast path for re-write. */
-		ret = ocfs2_get_block(inode, iblock, bh_result, create);
-
-		up_read(&oi->ip_alloc_sem);
-
+		ret = ocfs2_get_block_lock(inode, iblock, bh_result, create);
 		if (buffer_mapped(bh_result) &&
 		    !buffer_new(bh_result) &&
 		    ret == 0)
@@ -2424,7 +2433,7 @@ static ssize_t ocfs2_direct_IO(struct kiocb *iocb, struct iov_iter *iter)
 		return 0;

 	if (iov_iter_rw(iter) == READ)
-		get_block = ocfs2_get_block;
+		get_block = ocfs2_get_block_lock;
 	else
 		get_block = ocfs2_dio_get_block;

-- 1.9.5.msysgit.1

[Ocfs2-devel] [PATCH] ocfs2: the ip_alloc_sem should be taken in ocfs2_get_block()

[Changwei Ge]

Hi Alex,

Are you able to provide a way to reproduce this issue?
I'm very interested in it.

Thanks,
Changwei.

On 2017/10/20 17:08, alex chen wrote:
> The ip_alloc_sem should be taken in ocfs2_get_block() when reading file
> in DIRECT mode to prevent concurrent access to extent tree with
> ocfs2_dio_end_io_write(), which may cause BUGON in
> ocfs2_get_clusters_nocache()->BUG_ON(v_cluster < le32_to_cpu(rec->e_cpos))
> 
> Signed-off-by: Alex Chen <alex.chen@huawei.com>
> Reviewed-by: Jun Piao <piaojun@huawei.com>
> 
> ---
>   fs/ocfs2/aops.c | 21 +++++++++++++++------
>   1 file changed, 15 insertions(+), 6 deletions(-)
> 
> diff --git a/fs/ocfs2/aops.c b/fs/ocfs2/aops.c
> index 88a31e9..5cb939f 100644
> --- a/fs/ocfs2/aops.c
> +++ b/fs/ocfs2/aops.c
> @@ -134,6 +134,19 @@ static int ocfs2_symlink_get_block(struct inode *inode, sector_t iblock,
>   	return err;
>   }
> 
> +static int ocfs2_get_block_lock(struct inode *inode, sector_t iblock,
> +		    struct buffer_head *bh_result, int create)
> +{
> +	int ret;
> +	struct ocfs2_inode_info *oi = OCFS2_I(inode);
> +
> +	down_read(&oi->ip_alloc_sem);
> +	ret = ocfs2_get_block(inode, iblock, bh_result, create);
> +	up_read(&oi->ip_alloc_sem);
> +
> +	return ret;
> +}
> +
>   int ocfs2_get_block(struct inode *inode, sector_t iblock,
>   		    struct buffer_head *bh_result, int create)
>   {
> @@ -2154,12 +2167,8 @@ static int ocfs2_dio_get_block(struct inode *inode, sector_t iblock,
>   	 * while file size will be changed.
>   	 */
>   	if (pos + total_len <= i_size_read(inode)) {
> -		down_read(&oi->ip_alloc_sem);
>   		/* This is the fast path for re-write. */
> -		ret = ocfs2_get_block(inode, iblock, bh_result, create);
> -
> -		up_read(&oi->ip_alloc_sem);
> -
> +		ret = ocfs2_get_block_lock(inode, iblock, bh_result, create);
>   		if (buffer_mapped(bh_result) &&
>   		    !buffer_new(bh_result) &&
>   		    ret == 0)
> @@ -2424,7 +2433,7 @@ static ssize_t ocfs2_direct_IO(struct kiocb *iocb, struct iov_iter *iter)
>   		return 0;
> 
>   	if (iov_iter_rw(iter) == READ)
> -		get_block = ocfs2_get_block;
> +		get_block = ocfs2_get_block_lock;
>   	else
>   		get_block = ocfs2_dio_get_block;
> 
> -- 1.9.5.msysgit.1
> 
> 
> 
> _______________________________________________
> Ocfs2-devel mailing list
> Ocfs2-devel at oss.oracle.com
> https://oss.oracle.com/mailman/listinfo/ocfs2-devel
> 

[Ocfs2-devel] [PATCH] ocfs2: the ip_alloc_sem should be taken in ocfs2_get_block()

[Changwei Ge]

Hi,
Actually we encountered the same issue months ago and we also ever 
posted a patch to fix it, however, we never got a 'reviewed-by':-(

Your patch has a better change-log and the logic is fine to me.
I think it is deserved to be merged.

But it seems that your editor has '4 white spaces' occupied tab, it 
violates the exited code style. Can you adjust it to '8 white spaces' 
and re-send this patch?

Otherwise I thinks this is a good fix.

Moreover, this issue was introduced by commit c15471f79506 ("ocfs2: fix 
sparse file & data ordering issue in direct io")

Adding a *Fixs* tag to your change-log is encouraged.

Also I CC to ocfs2 maintainers so that they can provide some helpful 
comments.

Thanks,
Changwei.


On 2017/10/20 17:08, alex chen wrote:
> The ip_alloc_sem should be taken in ocfs2_get_block() when reading file
> in DIRECT mode to prevent concurrent access to extent tree with
> ocfs2_dio_end_io_write(), which may cause BUGON in
> ocfs2_get_clusters_nocache()->BUG_ON(v_cluster < le32_to_cpu(rec->e_cpos))
> 
> Signed-off-by: Alex Chen <alex.chen@huawei.com>
> Reviewed-by: Jun Piao <piaojun@huawei.com>

Acked-by: Changwei Ge <ge.changwei@h3c.com>

> 
> ---
>   fs/ocfs2/aops.c | 21 +++++++++++++++------
>   1 file changed, 15 insertions(+), 6 deletions(-)
> 
> diff --git a/fs/ocfs2/aops.c b/fs/ocfs2/aops.c
> index 88a31e9..5cb939f 100644
> --- a/fs/ocfs2/aops.c
> +++ b/fs/ocfs2/aops.c
> @@ -134,6 +134,19 @@ static int ocfs2_symlink_get_block(struct inode *inode, sector_t iblock,
>   	return err;
>   }
> 
> +static int ocfs2_get_block_lock(struct inode *inode, sector_t iblock,
> +		    struct buffer_head *bh_result, int create)
> +{
> +	int ret;
> +	struct ocfs2_inode_info *oi = OCFS2_I(inode);
> +
> +	down_read(&oi->ip_alloc_sem);
> +	ret = ocfs2_get_block(inode, iblock, bh_result, create);
> +	up_read(&oi->ip_alloc_sem);
> +
> +	return ret;
> +}
> +
>   int ocfs2_get_block(struct inode *inode, sector_t iblock,
>   		    struct buffer_head *bh_result, int create)
>   {
> @@ -2154,12 +2167,8 @@ static int ocfs2_dio_get_block(struct inode *inode, sector_t iblock,
>   	 * while file size will be changed.
>   	 */
>   	if (pos + total_len <= i_size_read(inode)) {
> -		down_read(&oi->ip_alloc_sem);
>   		/* This is the fast path for re-write. */
> -		ret = ocfs2_get_block(inode, iblock, bh_result, create);
> -
> -		up_read(&oi->ip_alloc_sem);
> -
> +		ret = ocfs2_get_block_lock(inode, iblock, bh_result, create);
>   		if (buffer_mapped(bh_result) &&
>   		    !buffer_new(bh_result) &&
>   		    ret == 0)
> @@ -2424,7 +2433,7 @@ static ssize_t ocfs2_direct_IO(struct kiocb *iocb, struct iov_iter *iter)
>   		return 0;
> 
>   	if (iov_iter_rw(iter) == READ)
> -		get_block = ocfs2_get_block;
> +		get_block = ocfs2_get_block_lock;
>   	else
>   		get_block = ocfs2_dio_get_block;
> 
> -- 1.9.5.msysgit.1
> 
> 
> 
> _______________________________________________
> Ocfs2-devel mailing list
> Ocfs2-devel at oss.oracle.com
> https://oss.oracle.com/mailman/listinfo/ocfs2-devel
> 

[Ocfs2-devel] [PATCH] ocfs2: the ip_alloc_sem should be taken in ocfs2_get_block()

[alex chen]

Hi Changwei,

On 2017/10/20 17:23, Changwei Ge wrote:
> Hi Alex,
> 
> Are you able to provide a way to reproduce this issue?
> I'm very interested in it.
> 

You can reproduce the BUG in the following steps:
1. touch "/mnt/ocfs2/test";
2. fallocate -l 1G "/mnt/ocfs2/test";
3. write the file "/mnt/ocfs2/test" using the io_submit();
4. at the same time, creating another process to read this file from
   a random location;

> Thanks,
> Changwei.
> 
> On 2017/10/20 17:08, alex chen wrote:
>> The ip_alloc_sem should be taken in ocfs2_get_block() when reading file
>> in DIRECT mode to prevent concurrent access to extent tree with
>> ocfs2_dio_end_io_write(), which may cause BUGON in
>> ocfs2_get_clusters_nocache()->BUG_ON(v_cluster < le32_to_cpu(rec->e_cpos))
>>
>> Signed-off-by: Alex Chen <alex.chen@huawei.com>
>> Reviewed-by: Jun Piao <piaojun@huawei.com>
>>
>> ---
>>   fs/ocfs2/aops.c | 21 +++++++++++++++------
>>   1 file changed, 15 insertions(+), 6 deletions(-)
>>
>> diff --git a/fs/ocfs2/aops.c b/fs/ocfs2/aops.c
>> index 88a31e9..5cb939f 100644
>> --- a/fs/ocfs2/aops.c
>> +++ b/fs/ocfs2/aops.c
>> @@ -134,6 +134,19 @@ static int ocfs2_symlink_get_block(struct inode *inode, sector_t iblock,
>>   	return err;
>>   }
>>
>> +static int ocfs2_get_block_lock(struct inode *inode, sector_t iblock,
>> +		    struct buffer_head *bh_result, int create)
>> +{
>> +	int ret;
>> +	struct ocfs2_inode_info *oi = OCFS2_I(inode);
>> +
>> +	down_read(&oi->ip_alloc_sem);
>> +	ret = ocfs2_get_block(inode, iblock, bh_result, create);
>> +	up_read(&oi->ip_alloc_sem);
>> +
>> +	return ret;
>> +}
>> +
>>   int ocfs2_get_block(struct inode *inode, sector_t iblock,
>>   		    struct buffer_head *bh_result, int create)
>>   {
>> @@ -2154,12 +2167,8 @@ static int ocfs2_dio_get_block(struct inode *inode, sector_t iblock,
>>   	 * while file size will be changed.
>>   	 */
>>   	if (pos + total_len <= i_size_read(inode)) {
>> -		down_read(&oi->ip_alloc_sem);
>>   		/* This is the fast path for re-write. */
>> -		ret = ocfs2_get_block(inode, iblock, bh_result, create);
>> -
>> -		up_read(&oi->ip_alloc_sem);
>> -
>> +		ret = ocfs2_get_block_lock(inode, iblock, bh_result, create);
>>   		if (buffer_mapped(bh_result) &&
>>   		    !buffer_new(bh_result) &&
>>   		    ret == 0)
>> @@ -2424,7 +2433,7 @@ static ssize_t ocfs2_direct_IO(struct kiocb *iocb, struct iov_iter *iter)
>>   		return 0;
>>
>>   	if (iov_iter_rw(iter) == READ)
>> -		get_block = ocfs2_get_block;
>> +		get_block = ocfs2_get_block_lock;
>>   	else
>>   		get_block = ocfs2_dio_get_block;
>>
>> -- 1.9.5.msysgit.1
>>
>>
>>
>> _______________________________________________
>> Ocfs2-devel mailing list
>> Ocfs2-devel at oss.oracle.com
>> https://oss.oracle.com/mailman/listinfo/ocfs2-devel
>>
> 
> 
> .
> 

[Ocfs2-devel] [PATCH] ocfs2: the ip_alloc_sem should be taken in ocfs2_get_block()

[alex chen]

Hi Changewei,

Thank for your suggestion, I will modify this patch and resend it.

Thanks
Alex

On 2017/10/21 11:27, Changwei Ge wrote:
> Hi,
> Actually we encountered the same issue months ago and we also ever 
> posted a patch to fix it, however, we never got a 'reviewed-by':-(
> 
> Your patch has a better change-log and the logic is fine to me.
> I think it is deserved to be merged.
> 
> But it seems that your editor has '4 white spaces' occupied tab, it 
> violates the exited code style. Can you adjust it to '8 white spaces' 
> and re-send this patch?
> 
> Otherwise I thinks this is a good fix.
> 
> Moreover, this issue was introduced by commit c15471f79506 ("ocfs2: fix 
> sparse file & data ordering issue in direct io")
> 
> Adding a *Fixs* tag to your change-log is encouraged.
> 
> Also I CC to ocfs2 maintainers so that they can provide some helpful 
> comments.
> 
> Thanks,
> Changwei.
> 
> 
> On 2017/10/20 17:08, alex chen wrote:
>> The ip_alloc_sem should be taken in ocfs2_get_block() when reading file
>> in DIRECT mode to prevent concurrent access to extent tree with
>> ocfs2_dio_end_io_write(), which may cause BUGON in
>> ocfs2_get_clusters_nocache()->BUG_ON(v_cluster < le32_to_cpu(rec->e_cpos))
>>
>> Signed-off-by: Alex Chen <alex.chen@huawei.com>
>> Reviewed-by: Jun Piao <piaojun@huawei.com>
> 
> Acked-by: Changwei Ge <ge.changwei@h3c.com>
> 
>>
>> ---
>>   fs/ocfs2/aops.c | 21 +++++++++++++++------
>>   1 file changed, 15 insertions(+), 6 deletions(-)
>>
>> diff --git a/fs/ocfs2/aops.c b/fs/ocfs2/aops.c
>> index 88a31e9..5cb939f 100644
>> --- a/fs/ocfs2/aops.c
>> +++ b/fs/ocfs2/aops.c
>> @@ -134,6 +134,19 @@ static int ocfs2_symlink_get_block(struct inode *inode, sector_t iblock,
>>   	return err;
>>   }
>>
>> +static int ocfs2_get_block_lock(struct inode *inode, sector_t iblock,
>> +		    struct buffer_head *bh_result, int create)
>> +{
>> +	int ret;
>> +	struct ocfs2_inode_info *oi = OCFS2_I(inode);
>> +
>> +	down_read(&oi->ip_alloc_sem);
>> +	ret = ocfs2_get_block(inode, iblock, bh_result, create);
>> +	up_read(&oi->ip_alloc_sem);
>> +
>> +	return ret;
>> +}
>> +
>>   int ocfs2_get_block(struct inode *inode, sector_t iblock,
>>   		    struct buffer_head *bh_result, int create)
>>   {
>> @@ -2154,12 +2167,8 @@ static int ocfs2_dio_get_block(struct inode *inode, sector_t iblock,
>>   	 * while file size will be changed.
>>   	 */
>>   	if (pos + total_len <= i_size_read(inode)) {
>> -		down_read(&oi->ip_alloc_sem);
>>   		/* This is the fast path for re-write. */
>> -		ret = ocfs2_get_block(inode, iblock, bh_result, create);
>> -
>> -		up_read(&oi->ip_alloc_sem);
>> -
>> +		ret = ocfs2_get_block_lock(inode, iblock, bh_result, create);
>>   		if (buffer_mapped(bh_result) &&
>>   		    !buffer_new(bh_result) &&
>>   		    ret == 0)
>> @@ -2424,7 +2433,7 @@ static ssize_t ocfs2_direct_IO(struct kiocb *iocb, struct iov_iter *iter)
>>   		return 0;
>>
>>   	if (iov_iter_rw(iter) == READ)
>> -		get_block = ocfs2_get_block;
>> +		get_block = ocfs2_get_block_lock;
>>   	else
>>   		get_block = ocfs2_dio_get_block;
>>
>> -- 1.9.5.msysgit.1
>>
>>
>>
>> _______________________________________________
>> Ocfs2-devel mailing list
>> Ocfs2-devel at oss.oracle.com
>> https://oss.oracle.com/mailman/listinfo/ocfs2-devel
>>
> 
> 
> .
> 

[Ocfs2-devel] [PATCH] ocfs2: the ip_alloc_sem should be taken in ocfs2_get_block()

[Joseph Qi]

On 17/10/20 17:03, alex chen wrote:
> The ip_alloc_sem should be taken in ocfs2_get_block() when reading file
> in DIRECT mode to prevent concurrent access to extent tree with
> ocfs2_dio_end_io_write(), which may cause BUGON in
> ocfs2_get_clusters_nocache()->BUG_ON(v_cluster < le32_to_cpu(rec->e_cpos))
> 
> Signed-off-by: Alex Chen <alex.chen@huawei.com>
> Reviewed-by: Jun Piao <piaojun@huawei.com>
> 
> ---
>  fs/ocfs2/aops.c | 21 +++++++++++++++------
>  1 file changed, 15 insertions(+), 6 deletions(-)
> 
> diff --git a/fs/ocfs2/aops.c b/fs/ocfs2/aops.c
> index 88a31e9..5cb939f 100644
> --- a/fs/ocfs2/aops.c
> +++ b/fs/ocfs2/aops.c
> @@ -134,6 +134,19 @@ static int ocfs2_symlink_get_block(struct inode *inode, sector_t iblock,
>  	return err;
>  }
> 
> +static int ocfs2_get_block_lock(struct inode *inode, sector_t iblock,
> +		    struct buffer_head *bh_result, int create)
> +{
> +	int ret;
> +	struct ocfs2_inode_info *oi = OCFS2_I(inode);
> +
> +	down_read(&oi->ip_alloc_sem);
> +	ret = ocfs2_get_block(inode, iblock, bh_result, create);
> +	up_read(&oi->ip_alloc_sem);
> +
> +	return ret;
> +}
> +
>  int ocfs2_get_block(struct inode *inode, sector_t iblock,
>  		    struct buffer_head *bh_result, int create)
>  {
> @@ -2154,12 +2167,8 @@ static int ocfs2_dio_get_block(struct inode *inode, sector_t iblock,
>  	 * while file size will be changed.
>  	 */
>  	if (pos + total_len <= i_size_read(inode)) {
> -		down_read(&oi->ip_alloc_sem);
>  		/* This is the fast path for re-write. */
> -		ret = ocfs2_get_block(inode, iblock, bh_result, create);
> -
> -		up_read(&oi->ip_alloc_sem);
> -
> +		ret = ocfs2_get_block_lock(inode, iblock, bh_result, create);
>  		if (buffer_mapped(bh_result) &&
>  		    !buffer_new(bh_result) &&
>  		    ret == 0)
> @@ -2424,7 +2433,7 @@ static ssize_t ocfs2_direct_IO(struct kiocb *iocb, struct iov_iter *iter)
>  		return 0;
> 
>  	if (iov_iter_rw(iter) == READ)
> -		get_block = ocfs2_get_block;
> +		get_block = ocfs2_get_block_lock;
ocfs2_lock_get_block may be better.

Thanks,
Joseph

>  	else
>  		get_block = ocfs2_dio_get_block;
> 
> -- 1.9.5.msysgit.1
> 
> 

[Ocfs2-devel] [PATCH] ocfs2: the ip_alloc_sem should be taken in ocfs2_get_block()

[Eric Ren]

Hi,

On 10/20/2017 05:03 PM, alex chen wrote:
> The ip_alloc_sem should be taken in ocfs2_get_block() when reading file
> in DIRECT mode to prevent concurrent access to extent tree with
> ocfs2_dio_end_io_write(), which may cause BUGON in
> ocfs2_get_clusters_nocache()->BUG_ON(v_cluster < le32_to_cpu(rec->e_cpos))

This maybe seem a obvious fix, but it would be great if you can
write a more detailed commit log, like paste the crash backtrace
here so that people can pick this fix easily when they see the same issue.

Thanks,
Eric
>
> Signed-off-by: Alex Chen <alex.chen@huawei.com>
> Reviewed-by: Jun Piao <piaojun@huawei.com>
>
> ---
>   fs/ocfs2/aops.c | 21 +++++++++++++++------
>   1 file changed, 15 insertions(+), 6 deletions(-)
>
> diff --git a/fs/ocfs2/aops.c b/fs/ocfs2/aops.c
> index 88a31e9..5cb939f 100644
> --- a/fs/ocfs2/aops.c
> +++ b/fs/ocfs2/aops.c
> @@ -134,6 +134,19 @@ static int ocfs2_symlink_get_block(struct inode *inode, sector_t iblock,
>   	return err;
>   }
>
> +static int ocfs2_get_block_lock(struct inode *inode, sector_t iblock,
> +		    struct buffer_head *bh_result, int create)
> +{
> +	int ret;
> +	struct ocfs2_inode_info *oi = OCFS2_I(inode);
> +
> +	down_read(&oi->ip_alloc_sem);
> +	ret = ocfs2_get_block(inode, iblock, bh_result, create);
> +	up_read(&oi->ip_alloc_sem);
> +
> +	return ret;
> +}
> +
>   int ocfs2_get_block(struct inode *inode, sector_t iblock,
>   		    struct buffer_head *bh_result, int create)
>   {
> @@ -2154,12 +2167,8 @@ static int ocfs2_dio_get_block(struct inode *inode, sector_t iblock,
>   	 * while file size will be changed.
>   	 */
>   	if (pos + total_len <= i_size_read(inode)) {
> -		down_read(&oi->ip_alloc_sem);
>   		/* This is the fast path for re-write. */
> -		ret = ocfs2_get_block(inode, iblock, bh_result, create);
> -
> -		up_read(&oi->ip_alloc_sem);
> -
> +		ret = ocfs2_get_block_lock(inode, iblock, bh_result, create);
>   		if (buffer_mapped(bh_result) &&
>   		    !buffer_new(bh_result) &&
>   		    ret == 0)
> @@ -2424,7 +2433,7 @@ static ssize_t ocfs2_direct_IO(struct kiocb *iocb, struct iov_iter *iter)
>   		return 0;
>
>   	if (iov_iter_rw(iter) == READ)
> -		get_block = ocfs2_get_block;
> +		get_block = ocfs2_get_block_lock;
>   	else
>   		get_block = ocfs2_dio_get_block;
>
> -- 1.9.5.msysgit.1
>
>
>

[Ocfs2-devel] [PATCH v2] ocfs2: the ip_alloc_sem should be taken in ocfs2_get_block()

[alex chen]

The ip_alloc_sem should be taken in ocfs2_get_block() when reading file
in DIRECT mode to prevent concurrent access to extent tree with
ocfs2_dio_end_io_write(), which may cause BUGON in
ocfs2_get_clusters_nocache()->BUG_ON(v_cluster < le32_to_cpu(rec->e_cpos))

Fixes: c15471f79506 ("ocfs2: fix sparse file & data ordering issue in direct io")

Signed-off-by: Alex Chen <alex.chen@huawei.com>
Reviewed-by: Jun Piao <piaojun@huawei.com>
Acked-by: Changwei Ge <ge.changwei@h3c.com>

---
 fs/ocfs2/aops.c | 22 ++++++++++++++++------
 1 file changed, 16 insertions(+), 6 deletions(-)

diff --git a/fs/ocfs2/aops.c b/fs/ocfs2/aops.c
index 88a31e9..29607d9 100644
--- a/fs/ocfs2/aops.c
+++ b/fs/ocfs2/aops.c
@@ -134,6 +134,19 @@ static int ocfs2_symlink_get_block(struct inode *inode, sector_t iblock,
 	return err;
 }

+static int ocfs2_get_block_lock(struct inode *inode, sector_t iblock,
+		    struct buffer_head *bh_result, int create)
+{
+	int ret = 0;
+	struct ocfs2_inode_info *oi = OCFS2_I(inode);
+
+	down_read(&oi->ip_alloc_sem);
+	ret = ocfs2_get_block(inode, iblock, bh_result, create);
+	up_read(&oi->ip_alloc_sem);
+
+	return ret;
+}
+
 int ocfs2_get_block(struct inode *inode, sector_t iblock,
 		    struct buffer_head *bh_result, int create)
 {
@@ -2154,12 +2167,9 @@ static int ocfs2_dio_get_block(struct inode *inode, sector_t iblock,
 	 * while file size will be changed.
 	 */
 	if (pos + total_len <= i_size_read(inode)) {
-		down_read(&oi->ip_alloc_sem);
-		/* This is the fast path for re-write. */
-		ret = ocfs2_get_block(inode, iblock, bh_result, create);
-
-		up_read(&oi->ip_alloc_sem);

+		/* This is the fast path for re-write. */
+		ret = ocfs2_get_block_lock(inode, iblock, bh_result, create);
 		if (buffer_mapped(bh_result) &&
 		    !buffer_new(bh_result) &&
 		    ret == 0)
@@ -2424,7 +2434,7 @@ static ssize_t ocfs2_direct_IO(struct kiocb *iocb, struct iov_iter *iter)
 		return 0;

 	if (iov_iter_rw(iter) == READ)
-		get_block = ocfs2_get_block;
+		get_block = ocfs2_get_block_lock;
 	else
 		get_block = ocfs2_dio_get_block;

-- 
1.9.5.msysgit.1

[Ocfs2-devel] [PATCH v2] ocfs2: the ip_alloc_sem should be taken in ocfs2_get_block()

[Gang He]

Hello Alex,

Base on your bug description, I think that the fix should be OK.
But I suggest you to use more clear function name in this patch.
1) For new-added function ocfs2_get_block_lock(), I think that it is better if the function name is ocfs2_dio_rd_get_block().
2) Then, rename the original function ocfs2_dio_get_block() to ocfs2_dio_wr_get_block().
Why? 
Since this function looks to only be used by direct IO (read) mode, 
we should make the function name more clear, do NOT make other developer to misunderstand/misuse it in the future.

Thanks
Gang 


>>> 
> The ip_alloc_sem should be taken in ocfs2_get_block() when reading file
> in DIRECT mode to prevent concurrent access to extent tree with
> ocfs2_dio_end_io_write(), which may cause BUGON in
> ocfs2_get_clusters_nocache()->BUG_ON(v_cluster < le32_to_cpu(rec->e_cpos))
> 
> Fixes: c15471f79506 ("ocfs2: fix sparse file & data ordering issue in direct 
> io")
> 
> Signed-off-by: Alex Chen <alex.chen@huawei.com>
> Reviewed-by: Jun Piao <piaojun@huawei.com>
> Acked-by: Changwei Ge <ge.changwei@h3c.com>
> 
> ---
>  fs/ocfs2/aops.c | 22 ++++++++++++++++------
>  1 file changed, 16 insertions(+), 6 deletions(-)
> 
> diff --git a/fs/ocfs2/aops.c b/fs/ocfs2/aops.c
> index 88a31e9..29607d9 100644
> --- a/fs/ocfs2/aops.c
> +++ b/fs/ocfs2/aops.c
> @@ -134,6 +134,19 @@ static int ocfs2_symlink_get_block(struct inode *inode, 
> sector_t iblock,
>  	return err;
>  }
> 
> +static int ocfs2_get_block_lock(struct inode *inode, sector_t iblock,
> +		    struct buffer_head *bh_result, int create)
> +{
> +	int ret = 0;
> +	struct ocfs2_inode_info *oi = OCFS2_I(inode);
> +
> +	down_read(&oi->ip_alloc_sem);
> +	ret = ocfs2_get_block(inode, iblock, bh_result, create);
> +	up_read(&oi->ip_alloc_sem);
> +
> +	return ret;
> +}
> +
>  int ocfs2_get_block(struct inode *inode, sector_t iblock,
>  		    struct buffer_head *bh_result, int create)
>  {
> @@ -2154,12 +2167,9 @@ static int ocfs2_dio_get_block(struct inode *inode, 
> sector_t iblock,
>  	 * while file size will be changed.
>  	 */
>  	if (pos + total_len <= i_size_read(inode)) {
> -		down_read(&oi->ip_alloc_sem);
> -		/* This is the fast path for re-write. */
> -		ret = ocfs2_get_block(inode, iblock, bh_result, create);
> -
> -		up_read(&oi->ip_alloc_sem);
> 
> +		/* This is the fast path for re-write. */
> +		ret = ocfs2_get_block_lock(inode, iblock, bh_result, create);
>  		if (buffer_mapped(bh_result) &&
>  		    !buffer_new(bh_result) &&
>  		    ret == 0)
> @@ -2424,7 +2434,7 @@ static ssize_t ocfs2_direct_IO(struct kiocb *iocb, 
> struct iov_iter *iter)
>  		return 0;
> 
>  	if (iov_iter_rw(iter) == READ)
> -		get_block = ocfs2_get_block;
> +		get_block = ocfs2_get_block_lock;
>  	else
>  		get_block = ocfs2_dio_get_block;
> 
> -- 
> 1.9.5.msysgit.1
> 
> 
> _______________________________________________
> Ocfs2-devel mailing list
> Ocfs2-devel at oss.oracle.com 
> https://oss.oracle.com/mailman/listinfo/ocfs2-devel

[Ocfs2-devel] [PATCH v2] ocfs2: the ip_alloc_sem should be taken in ocfs2_get_block()

[alex chen]

Hi Everybody,

Thanks for your reviews.

I agree with Gang to make function name ocfs2_get_block_lock() more
readable. But the ocfs2_get_block_lock() may be called by the
ocfs2_dio_get_block() in re-write sitiation, so I don't think the
ocfs2_dio_rd_get_block() is a better name. I agree with Joseph and
rename it to ocfs2_lock_get_block().

BTW, I think the ocfs2_dio_wr_get_block() is more suitable for the
ocfs2_dio_get_block().

Please give me any feedback if you have any suggestions.

Thanks
Alex

On 2017/10/23 15:16, Gang He wrote:
> Hello Alex,
> 
> Base on your bug description, I think that the fix should be OK.
> But I suggest you to use more clear function name in this patch.
> 1) For new-added function ocfs2_get_block_lock(), I think that it is better if the function name is ocfs2_dio_rd_get_block().
> 2) Then, rename the original function ocfs2_dio_get_block() to ocfs2_dio_wr_get_block().
> Why? 
> Since this function looks to only be used by direct IO (read) mode, 
> we should make the function name more clear, do NOT make other developer to misunderstand/misuse it in the future.
> 
> Thanks
> Gang 
> 
> 
>>>>
>> The ip_alloc_sem should be taken in ocfs2_get_block() when reading file
>> in DIRECT mode to prevent concurrent access to extent tree with
>> ocfs2_dio_end_io_write(), which may cause BUGON in
>> ocfs2_get_clusters_nocache()->BUG_ON(v_cluster < le32_to_cpu(rec->e_cpos))
>>
>> Fixes: c15471f79506 ("ocfs2: fix sparse file & data ordering issue in direct 
>> io")
>>
>> Signed-off-by: Alex Chen <alex.chen@huawei.com>
>> Reviewed-by: Jun Piao <piaojun@huawei.com>
>> Acked-by: Changwei Ge <ge.changwei@h3c.com>
>>
>> ---
>>  fs/ocfs2/aops.c | 22 ++++++++++++++++------
>>  1 file changed, 16 insertions(+), 6 deletions(-)
>>
>> diff --git a/fs/ocfs2/aops.c b/fs/ocfs2/aops.c
>> index 88a31e9..29607d9 100644
>> --- a/fs/ocfs2/aops.c
>> +++ b/fs/ocfs2/aops.c
>> @@ -134,6 +134,19 @@ static int ocfs2_symlink_get_block(struct inode *inode, 
>> sector_t iblock,
>>  	return err;
>>  }
>>
>> +static int ocfs2_get_block_lock(struct inode *inode, sector_t iblock,
>> +		    struct buffer_head *bh_result, int create)
>> +{
>> +	int ret = 0;
>> +	struct ocfs2_inode_info *oi = OCFS2_I(inode);
>> +
>> +	down_read(&oi->ip_alloc_sem);
>> +	ret = ocfs2_get_block(inode, iblock, bh_result, create);
>> +	up_read(&oi->ip_alloc_sem);
>> +
>> +	return ret;
>> +}
>> +
>>  int ocfs2_get_block(struct inode *inode, sector_t iblock,
>>  		    struct buffer_head *bh_result, int create)
>>  {
>> @@ -2154,12 +2167,9 @@ static int ocfs2_dio_get_block(struct inode *inode, 
>> sector_t iblock,
>>  	 * while file size will be changed.
>>  	 */
>>  	if (pos + total_len <= i_size_read(inode)) {
>> -		down_read(&oi->ip_alloc_sem);
>> -		/* This is the fast path for re-write. */
>> -		ret = ocfs2_get_block(inode, iblock, bh_result, create);
>> -
>> -		up_read(&oi->ip_alloc_sem);
>>
>> +		/* This is the fast path for re-write. */
>> +		ret = ocfs2_get_block_lock(inode, iblock, bh_result, create);
>>  		if (buffer_mapped(bh_result) &&
>>  		    !buffer_new(bh_result) &&
>>  		    ret == 0)
>> @@ -2424,7 +2434,7 @@ static ssize_t ocfs2_direct_IO(struct kiocb *iocb, 
>> struct iov_iter *iter)
>>  		return 0;
>>
>>  	if (iov_iter_rw(iter) == READ)
>> -		get_block = ocfs2_get_block;
>> +		get_block = ocfs2_get_block_lock;
>>  	else
>>  		get_block = ocfs2_dio_get_block;
>>
>> -- 
>> 1.9.5.msysgit.1
>>
>>
>> _______________________________________________
>> Ocfs2-devel mailing list
>> Ocfs2-devel at oss.oracle.com 
>> https://oss.oracle.com/mailman/listinfo/ocfs2-devel
> 
> 
> .
>