97533 - UMRs in InitGlobals

Status: RESOLVED FIXED

(Core :: Layout, defect)

Description

[David Baron :dbaron: (⌚️UTC-4, no longer working on Mozilla)]

There are a bunch of UMRs (uninitialized memory reads) in InitGlobals in
nsFontMetricsGTK.cpp.  I'll attach a patch that does what I think you meant to
do, but you'll have to check that it really is what you meant.

This should be a high priority to fix since UMRs can result in random behavior,
particularly with newer compilers and at higher optimization levels.

Comment 1

[David Baron :dbaron: (⌚️UTC-4, no longer working on Mozilla)]

Attachment: proposed patch

Comment 2

[kill this account]

Hum... How did I, the reviewer, and the super-reviewer miss these?

The patch looks good.

r=[e-mail]

Comment 3

[kill this account]

One thing to note: 

The uninitialized status value may cause the program to incorrectly choose
the fallback, the default, or the user pref setting.

In any case, the values are being set to an initialized (not random) value.

Comment 4

[Frank Tang]

I think we should try to land this on m9.4
please ask sr and a= 

Comment 5

[Roland Mainz]

Mhhh... it looks that this was fixed in Xlib-toolkit. Should I file a bug to
port other improvements from Xlib-toolkit back to GTK+ toolkit ?

Comment 6

[Frank Tang]

>Mhhh... it looks that this was fixed in Xlib-toolkit. Should I file a bug to
>port other improvements from Xlib-toolkit back to GTK+ toolkit ?
hum???? this is a mozila bug. 

ok, this is not ramdon, then move it to m0.9.5 

Comment 7

[David Baron :dbaron: (⌚️UTC-4, no longer working on Mozilla)]

Binary randomness is still randomness, and still very bad.

Comment 8

[Christopher Blizzard (:blizzard)]

Comment on attachment 47571 [details] [diff] [review]
proposed patch

sr=blizzard

Comment 9

[Frank Tang]

bstell said it is radomly pick between the default and user preference. I think
it is still bad enough to ask for a 9.4 check in. move to m0.9.4

Comment 10

[Jaime Rodriguez, Jr.]

If this makes it into 0.9.4, we'll take it for nsbranch.

Comment 11

[kill this account]

checked into trunk

Comment 12

[Asa Dotzler [:asa]]

Comment on attachment 47571 [details] [diff] [review]
proposed patch

a=asa on behalf of drivers for checkin to 0.9.4 branch

Comment 13

[kill this account]

checked in 0.9.4 branch

Comment 14

[Chris Petersen]

Not sure how QA can verify this. Marking verified per last comments in the Sept
25th branch build.