1338498 - Incorrect size of memset in vp8_create_decoder_instances

Status: RESOLVED DUPLICATE of bug 1480374

(Core :: WebRTC: Audio/Video, defect, P5)

Description

[Julian Seward [:jseward]]

gcc 7 -Wall reports

Warning: -Wmemset-elt-size in /home/sewardj/MOZ/MC-TALL/media/libvpx/libvpx/vp8/decoder/onyxd_if.c: ‘memset’ used with length equal to number of elements without multiplication by element size

home/sewardj/MOZ/MC-TALL/media/libvpx/libvpx/vp8/decoder/onyxd_if.c:449:5: warning: ‘memset’ used with length equal to number of elements without multiplication by element size [-Wmemset-elt-size]
      memset(fb->pbi, 0, sizeof(fb->pbi) / sizeof(fb->pbi[0]));

Comment 1

[Julian Seward [:jseward]]

Attachment: bug1338498-1.cset

Comment 2

[Randell Jesup [:jesup] (needinfo me)]

Comment on attachment 8835979 [details] [diff] [review]
bug1338498-1.cset

Review of attachment 8835979 [details] [diff] [review]:
-----------------------------------------------------------------

punting review over to a libvpx maintainer who imported 1.6.1

::: media/libvpx/libvpx/vp8/decoder/onyxd_if.c
@@ +445,5 @@
>  
>  #if CONFIG_MULTITHREAD
>    if (setjmp(fb->pbi[0]->common.error.jmp)) {
>      vp8_remove_decoder_instances(fb);
> +    memset(fb->pbi, 0, sizeof(fb->pbi));

I think this doesn't intend to clear all of pbi, but only the first element of it (given the code around it with all the hard-coded [0]'s, and the same in the if()).

Comment 3

[johannkoenig]

https://bugs.chromium.org/p/webm/issues/detail?id=1364

Comment 4

[Julian Seward [:jseward]]

Johann, would it be possible to resolve this?

Comment 5

[Julian Seward [:jseward]]

Johann, ping?

Comment 6

[James Zern]

I submitted the fix for this:
v1.6.1-284-g47cf7c25a
47cf7c25a Merge "vp8_create_decoder_instances: correct pbi[] memset"
9c3c1f372 vp8_create_decoder_instances: correct pbi[] memset

Comment 7

[Randell Jesup [:jesup] (needinfo me)]

Comment on attachment 8835979 [details] [diff] [review]
bug1338498-1.cset

Review of attachment 8835979 [details] [diff] [review]:
-----------------------------------------------------------------

Exact same patch landed upstream, r+'ing on that basis

Comment 8

[Randell Jesup [:jesup] (needinfo me)]

jseward: NOTE: patch is not landable as-is.  A .patch file must be added to the libvpx directory, and the import script (update.py) updated to apply it.

Comment 9

[Bulk Bug Changes for mreavy's org]

Mass change P4->P5 to align with new Mozilla triage process.

Comment 10

[Daniel Veditz [:dveditz]]

Duping forward because the patch has landed in that bug.