1131250 - [Camera][Gecko] MapIsoFromGonk() SIGSEGVs on aIso == nullptr

Status: RESOLVED FIXED

(Firefox OS Graveyard :: Gaia::Camera, defect)

Description

[Mike Habicher [:mikeh] (high bugzilla latency)]

Found while investigating camera on Shinano.

Comment 1

[Mike Habicher [:mikeh] (high bugzilla latency)]

Attachment: Handle ISO_MODE value == nullptr, v1

Comment 2

[Andrew Osmond [:aosmond] (he/him)]

Comment on attachment 8561611 [details] [diff] [review]
Handle ISO_MODE value == nullptr, v1

Review of attachment 8561611 [details] [diff] [review]:
-----------------------------------------------------------------

LGTM although I note that we print out the potentially null string in GonkCameraParameters::Initialize. I assume that all of our various printf paths are %s + null pointer safe?

Comment 3

[Mike Habicher [:mikeh] (high bugzilla latency)]

(In reply to Andrew Osmond [:aosmond] from comment #2)

> LGTM although I note that we print out the potentially null string in
> GonkCameraParameters::Initialize. I assume that all of our various printf
> paths are %s + null pointer safe?

In the Initialize() case, the array of isoModes has a zero length, so nothing gets printed.

Right now, this bug really only pops up when inspecting the object in the WebIDE, where .isoMode gets queried regardless of what capabilities.isoModes returns. So this just makes it more robust.

Comment 4

[Mike Habicher [:mikeh] (high bugzilla latency)]

Attachment: Handle ISO_MODE value == nullptr, v2

Make the calling function more robust as well. Practically, this means we'll never actually hit MapIsoFromGonk() at all, but we should keep the !aIso check in there anyway just in case.

Comment 5

[Mike Habicher [:mikeh] (high bugzilla latency)]

https://hg.mozilla.org/integration/b2g-inbound/rev/57362968b46a

Comment 6

[Ryan VanderMeulen [:RyanVM]]

https://hg.mozilla.org/mozilla-central/rev/57362968b46a