[PATCH] ubi: fastmap: Don't produce the initial anchor PEB when fastmap is disabled

[PATCH] ubi: fastmap: Don't produce the initial anchor PEB when fastmap is disabled

[Zhihao Cheng]

Following process triggers a memleak caused by forgetting to release the
initial anchor PEB (CONFIG_MTD_UBI_FASTMAP is disabled):
1. attach -> __erase_worker -> produce the initial anchor PEB
2. detach -> ubi_fastmap_close (Do nothing, it should have released the
   initial anchor PEB)

Don't produce the initial anchor PEB in __erase_worker() when fastmap
is disabled.

Signed-off-by: Zhihao Cheng <chengzhihao1@huawei.com>
Fixes: f9c34bb529975fe ("ubi: Fix producing anchor PEBs")
Reported-by: syzbot+d9aab50b1154e3d163f5@syzkaller.appspotmail.com
---
 drivers/mtd/ubi/wl.c | 8 +++++++-
 1 file changed, 7 insertions(+), 1 deletion(-)

diff --git a/drivers/mtd/ubi/wl.c b/drivers/mtd/ubi/wl.c
index 5146cce5fe32..5ebe1084a8e7 100644
--- a/drivers/mtd/ubi/wl.c
+++ b/drivers/mtd/ubi/wl.c
@@ -1079,13 +1079,19 @@ static int __erase_worker(struct ubi_device *ubi, struct ubi_work *wl_wrk)
 	if (!err) {
 		spin_lock(&ubi->wl_lock);
 
-		if (!ubi->fm_anchor && e->pnum < UBI_FM_MAX_START) {
+#ifdef CONFIG_MTD_UBI_FASTMAP
+		if (!ubi->fm_disabled && !ubi->fm_anchor &&
+		    e->pnum < UBI_FM_MAX_START) {
 			ubi->fm_anchor = e;
 			ubi->fm_do_produce_anchor = 0;
 		} else {
 			wl_tree_add(e, &ubi->free);
 			ubi->free_count++;
 		}
+#else
+		wl_tree_add(e, &ubi->free);
+		ubi->free_count++;
+#endif
 
 		spin_unlock(&ubi->wl_lock);
 
-- 
2.25.4

Re: [PATCH] ubi: fastmap: Don't produce the initial anchor PEB when fastmap is disabled

[Sascha Hauer]

Hi,

On Mon, Jun 01, 2020 at 05:11:34PM +0800, Zhihao Cheng wrote:
> Following process triggers a memleak caused by forgetting to release the
> initial anchor PEB (CONFIG_MTD_UBI_FASTMAP is disabled):
> 1. attach -> __erase_worker -> produce the initial anchor PEB
> 2. detach -> ubi_fastmap_close (Do nothing, it should have released the
>    initial anchor PEB)
> 
> Don't produce the initial anchor PEB in __erase_worker() when fastmap
> is disabled.
> 
> Signed-off-by: Zhihao Cheng <chengzhihao1@huawei.com>
> Fixes: f9c34bb529975fe ("ubi: Fix producing anchor PEBs")
> Reported-by: syzbot+d9aab50b1154e3d163f5@syzkaller.appspotmail.com
> ---
>  drivers/mtd/ubi/wl.c | 8 +++++++-
>  1 file changed, 7 insertions(+), 1 deletion(-)
> 
> diff --git a/drivers/mtd/ubi/wl.c b/drivers/mtd/ubi/wl.c
> index 5146cce5fe32..5ebe1084a8e7 100644
> --- a/drivers/mtd/ubi/wl.c
> +++ b/drivers/mtd/ubi/wl.c
> @@ -1079,13 +1079,19 @@ static int __erase_worker(struct ubi_device *ubi, struct ubi_work *wl_wrk)
>  	if (!err) {
>  		spin_lock(&ubi->wl_lock);
>  
> -		if (!ubi->fm_anchor && e->pnum < UBI_FM_MAX_START) {
> +#ifdef CONFIG_MTD_UBI_FASTMAP
> +		if (!ubi->fm_disabled && !ubi->fm_anchor &&
> +		    e->pnum < UBI_FM_MAX_START) {

Rather than introducing another #ifdef you could do a

		if (IS_ENABLED(CONFIG_MTD_UBI_FASTMAP) &&
		    !ubi->fm_disabled && !ubi->fm_anchor &&
		    e->pnum < UBI_FM_MAX_START)

And I am not sure if the IS_ENABLED(CONFIG_MTD_UBI_FASTMAP) is necessary
at all because we do a ubi->fm_disabled = 1 when fastmap is disabled.

Regards,
 Sascha

-- 
Pengutronix e.K.                           |                             |
Steuerwalder Str. 21                       | http://www.pengutronix.de/  |
31137 Hildesheim, Germany                  | Phone: +49-5121-206917-0    |
Amtsgericht Hildesheim, HRA 2686           | Fax:   +49-5121-206917-5555 |

Re: [PATCH] ubi: fastmap: Don't produce the initial anchor PEB when fastmap is disabled

[Zhihao Cheng]

在 2020/6/2 17:23, Sascha Hauer 写道:
> Hi,
>
> On Mon, Jun 01, 2020 at 05:11:34PM +0800, Zhihao Cheng wrote:
>> Following process triggers a memleak caused by forgetting to release the
>> initial anchor PEB (CONFIG_MTD_UBI_FASTMAP is disabled):
>> 1. attach -> __erase_worker -> produce the initial anchor PEB
>> 2. detach -> ubi_fastmap_close (Do nothing, it should have released the
>>     initial anchor PEB)
>>
>> Don't produce the initial anchor PEB in __erase_worker() when fastmap
>> is disabled.
>>
>> Signed-off-by: Zhihao Cheng <chengzhihao1@huawei.com>
>> Fixes: f9c34bb529975fe ("ubi: Fix producing anchor PEBs")
>> Reported-by: syzbot+d9aab50b1154e3d163f5@syzkaller.appspotmail.com
>> ---
>>   drivers/mtd/ubi/wl.c | 8 +++++++-
>>   1 file changed, 7 insertions(+), 1 deletion(-)
>>
>> diff --git a/drivers/mtd/ubi/wl.c b/drivers/mtd/ubi/wl.c
>> index 5146cce5fe32..5ebe1084a8e7 100644
>> --- a/drivers/mtd/ubi/wl.c
>> +++ b/drivers/mtd/ubi/wl.c
>> @@ -1079,13 +1079,19 @@ static int __erase_worker(struct ubi_device *ubi, struct ubi_work *wl_wrk)
>>   	if (!err) {
>>   		spin_lock(&ubi->wl_lock);
>>   
>> -		if (!ubi->fm_anchor && e->pnum < UBI_FM_MAX_START) {
>> +#ifdef CONFIG_MTD_UBI_FASTMAP
>> +		if (!ubi->fm_disabled && !ubi->fm_anchor &&
>> +		    e->pnum < UBI_FM_MAX_START) {
> Rather than introducing another #ifdef you could do a
>
> 		if (IS_ENABLED(CONFIG_MTD_UBI_FASTMAP) &&
> 		    !ubi->fm_disabled && !ubi->fm_anchor &&
> 		    e->pnum < UBI_FM_MAX_START)
>
> And I am not sure if the IS_ENABLED(CONFIG_MTD_UBI_FASTMAP) is necessary
> at all because we do a ubi->fm_disabled = 1 when fastmap is disabled.
>
> Regards,
>   Sascha
>
Agree.