Re: [PATCH V3] bcache: revert replacing IS_ERR_OR_NULL with IS_ERR again

[邹明哲 <mingzhe.zou@xxxxxxxxxxxx>]

LGTM

reviewed by Mingzhe Zou

Original:
From：Coly Li <colyli@xxxxxxx>
Date：2024-11-27 11:17:01(中国 (GMT+08:00))
To：liequan che <liequanche@xxxxxxxxx>
Cc：mingzhe.zou@xxxxxxxxxxxx<mingzhe.zou@xxxxxxxxxxxx> , Kent Overstreet <kent.overstreet@xxxxxxxxx> , linux-bcache <linux-bcache@xxxxxxxxxxxxxxx> , linux-kernel <linux-kernel@xxxxxxxxxxxxxxx>
Subject：Re: [PATCH V3] bcache: revert replacing IS_ERR_OR_NULL with IS_ERR again
&gt; 2024年11月27日 11:01，liequan che <liequanche@xxxxxxxxx> 写道：
&gt; 
&gt; Commit 028ddcac477b ("bcache: Remove unnecessary NULL point check in
&gt; node allocations") leads a NULL pointer deference in cache_set_flush().
&gt; 
&gt; 1721         if (!IS_ERR_OR_NULL(c-&gt;root))
&gt; 1722                 list_add(&amp;c-&gt;root-&gt;list, &amp;c-&gt;btree_cache);
&gt; 
&gt; From the above code in cache_set_flush(), if previous registration code
&gt; fails before allocating c-&gt;root, it is possible c-&gt;root is NULL as what
&gt; it is initialized. Also __bch_btree_node_alloc() never returns NULL but
&gt; c-&gt;root is possible to be NULL at above line 1721.
&gt; 
&gt; This patch replaces IS_ERR() by IS_ERR_OR_NULL() to fix this.


OK, this time the commit log makes a lot sense. It is clear to me.


&gt; 
&gt; Fixes: 028ddcac477b ("bcache: Remove unnecessary NULL point check in
&gt; node allocations")
&gt; Signed-off-by: Liequan Che <cheliequan@xxxxxxxxxx>
&gt; Cc: stable@xxxxxxxxxxxxxxx
&gt; Cc: Zheng Wang <zyytlz.wz@xxxxxxx>
&gt; Cc: Coly Li <colyli@xxxxxxx>
&gt; ---
&gt; drivers/md/bcache/super.c | 2 +-
&gt; 1 file changed, 1 insertion(+), 1 deletion(-)
&gt; 
&gt; diff --git a/drivers/md/bcache/super.c b/drivers/md/bcache/super.c
&gt; index e7abfdd77c3b..e42f1400cea9 100644
&gt; --- a/drivers/md/bcache/super.c
&gt; +++ b/drivers/md/bcache/super.c
&gt; @@ -1718,7 +1718,7 @@ static CLOSURE_CALLBACK(cache_set_flush)
&gt;        if (!IS_ERR_OR_NULL(c-&gt;gc_thread))
&gt;                kthread_stop(c-&gt;gc_thread);
&gt; 
&gt; -       if (!IS_ERR(c-&gt;root))
&gt; +       if (!IS_ERR_OR_NULL(c-&gt;root))
&gt;                list_add(&amp;c-&gt;root-&gt;list, &amp;c-&gt;btree_cache);
&gt; 
&gt;        /*
&gt; —
&gt; 2.46.0

It is fine to me. I am in travel these days, and will handle this patch by end of this week.

Thanks for composing this patch.

Coly Li
</colyli@xxxxxxx></zyytlz.wz@xxxxxxx></cheliequan@xxxxxxxxxx></liequanche@xxxxxxxxx></linux-kernel@xxxxxxxxxxxxxxx></linux-bcache@xxxxxxxxxxxxxxx></kent.overstreet@xxxxxxxxx></mingzhe.zou@xxxxxxxxxxxx></liequanche@xxxxxxxxx></colyli@xxxxxxx>