Re: ALSA: control - add layer registration routines

Dne 31. 03. 21 v 17:17 Colin Ian King napsal(a):

Hi,

Static analysis on linux-next with Coverity has detected a potential issue in the following commit:

commit 3f0638a0333bfdd0549985aa620f2ab69737af47 Author: Jaroslav Kysela perex@perex.cz Date: Wed Mar 17 18:29:41 2021 +0100

ALSA: control - add layer registration routines

The static analysis is as follows:

2072 void snd_ctl_disconnect_layer(struct snd_ctl_layer_ops *lops) 2073 { 2074 struct snd_ctl_layer_ops *lops2, *prev_lops2; 2075 2076 down_write(&snd_ctl_layer_rwsem);

assignment: Assigning: prev_lops2 = NULL.

2077 for (lops2 = snd_ctl_layer, prev_lops2 = NULL; lops2; lops2 = lops2->next) 2078 if (lops2 == lops) {

null: At condition prev_lops2, the value of prev_lops2 must be NULL.
dead_error_condition: The condition !prev_lops2 must be true.

2079 if (!prev_lops2) 2080 snd_ctl_layer = lops->next; 2081 else

'Constant' variable guards dead code (DEADCODE) dead_error_line:
Execution cannot reach this statement: prev_lops2->next = lops->next;.
Local variable prev_lops2 is assigned only once, to a constant

value, making it effectively constant throughout its scope. If this is not the intent, examine the logic to see if there is a missing assignment that would make prev_lops2 not remain constant.

2082 prev_lops2->next = lops->next; 2083 break; 2084 } 2085 up_write(&snd_ctl_layer_rwsem); 2086 }

I couldn't quite figure out the original intent of the prev_lops use, so I'd thought I'd report this issue as the code does look incorrect.

Thank you. I submitted the fix here:

https://lore.kernel.org/alsa-devel/20210331180702.663489-1-perex@perex.cz/

Jaroslav