Open
Bug 543556
Opened 16 years ago
Updated 3 years ago
NSPR tests fail on Android due to hardcoded "/tmp" paths
Categories
(NSPR :: NSPR, defect)
Tracking
(Not tracked)
NEW
People
(Reporter: bcombee, Unassigned)
References
Details
Attachments
(2 files)
|
580 bytes,
patch
|
ted
:
review+
wtc
:
superreview+
|
Details | Diff | Splinter Review |
|
4.68 KB,
patch
|
ted
:
review-
|
Details | Diff | Splinter Review |
Android devices do not have a "/tmp" folder. Some of the NSPR tests assume that tmp exists and is writable.
For a RAM-based filesystem, you can create a /dev/tmp folder and put files there. You also can just create a folder under /data and put files there or use the current directory like many tests do.
The failing tests because of no /tmp are:
append.c
foreign.c
sema.c
semaerr1.c
semaerr.c
semaping.c/semapong.c
socket.c
testfile.c
| Reporter | ||
Comment 1•16 years ago
|
||
Need this patch applied before you can even build these tests
Depends on: 542113
| Reporter | ||
Comment 2•16 years ago
|
||
even with /tmp location fixed, I'm finding semaphore tests failing on Android. The other tests I listed pass when they use files in the current directory rather than in /tmp.
Comment 3•16 years ago
|
||
Ben: you can ignore the semaphore tests if your goal is to
port Mozilla/Firefox. Mozilla/Firefox does not use NSPR
interprocess semaphores.
| Reporter | ||
Comment 4•16 years ago
|
||
Thanks, wtc! That's a bit less debugging/patching that I'll have to do.
| Reporter | ||
Comment 5•16 years ago
|
||
socket test also failed because of missing /tmp, this fixes it
| Reporter | ||
Comment 6•16 years ago
|
||
Comment 7•16 years ago
|
||
Comment on attachment 426049 [details] [diff] [review]
This is the first set of changes not including socket
Wan-Teh, do you have any comments on these Ben's patches?
Attachment #426049 -
Flags: review?(wtc)
Updated•16 years ago
|
Attachment #426048 -
Flags: review?(wtc)
Updated•16 years ago
|
Attachment #426048 -
Flags: review?(wtc) → review?(ted.mielczarek)
Updated•16 years ago
|
Attachment #426049 -
Flags: review?(wtc) → review?(ted.mielczarek)
Updated•16 years ago
|
Attachment #426048 -
Flags: review?(ted.mielczarek) → review?(wtc)
Updated•16 years ago
|
Attachment #426049 -
Flags: review?(ted.mielczarek) → review?(wtc)
Updated•16 years ago
|
Attachment #426049 -
Flags: review?(wtc) → review?(ted.mielczarek)
Updated•16 years ago
|
Attachment #426048 -
Flags: review?(wtc) → review?(ted.mielczarek)
Updated•16 years ago
|
Assignee: wtc → ben.combee
Comment 8•16 years ago
|
||
Comment on attachment 426048 [details] [diff] [review]
Fix for /tmp use in socket test
Please use #elif defined(ANDROID) for consistency.
Attachment #426048 -
Flags: review?(ted.mielczarek) → review+
Comment 9•16 years ago
|
||
Comment on attachment 426049 [details] [diff] [review]
This is the first set of changes not including socket
I think you should consistently use /data as the temp dir instead of using relative pathnames.
Attachment #426049 -
Flags: review?(ted.mielczarek) → review-
Comment 10•16 years ago
|
||
Comment on attachment 426048 [details] [diff] [review]
Fix for /tmp use in socket test
r=wtc. Please make the following change before you check this in.
>+#elif ANDROID
The NSPR style is to test if the macro is defined:
+#elif defined(ANDROID)
Attachment #426048 -
Flags: superreview+
Comment 11•4 years ago
|
||
The bug assignee is inactive on Bugzilla, so the assignee is being reset.
Assignee: ben.combee → nobody
Updated•3 years ago
|
Severity: normal → S3
You need to log in
before you can comment on or make changes to this bug.
Description
•