Closed
Bug 30002
Opened 25 years ago
Closed 25 years ago
Bugzilla query page makes lots of read/write/select calls
Categories
(Core Graveyard :: Tracking, defect, P3)
Tracking
(Not tracked)
VERIFIED
FIXED
People
(Reporter: tomi.leppikangas, Assigned: pavlov)
References
()
Details
(Keywords: perf)
Attachments
(2 files)
Bugzilla query page makes lots of read/write/select system calls.
This is shown by running "strace dist/bin/mozilla-bin"
At end of page loading and before page is displayed, there is many
read/write/select calls:
select(15, [14], NULL, NULL, NULL) = 1 (in [14])
read(14, "\1\0011\257\0\0\0\0\213H@\0\10\1t\0\0\0\0\0\0\0\0\0\32"..., 32) = 32
write(14, "(\0\4\0o\f\300\5%\0\0\0\0\0\0\0", 16) = 16
read(14, 0xbfffc124, 32) = -1 EAGAIN (Resource temporarily unavailable)
select(15, [14], NULL, NULL, NULL) = 1 (in [14])
read(14, "\1\0012\257\0\0\0\0\213H@\0\10\1t\0\0\0\0\0\1\0\0\0\32"..., 32) = 32
write(14, "\f\0\5\0\306\r\300\5\3\0\0\0H\1\0\0\26\4\0\0(\0\4\0o\f"..., 36) = 36
read(14, 0xbfffc8f8, 32) = -1 EAGAIN (Resource temporarily unavailable)
select(15, [14], NULL, NULL, NULL) = 1 (in [14])
.
.
.
I am not sure is this performance bug, but seems that there is too
many of those. There is bug #29544 for other query page perf problem.
Same thing happens on debug and opt build, and nightlyes too.
Reporter | ||
Comment 1•25 years ago
|
||
I looked a bit more of this, seems that the problem is combo boxes.
I made two testpages, both have 30 select widgets. Other have size=1, other
size=4. Page with size=4 loads in 3sec, and page with size=1 loads in 40sec.
Page with size=1 makes zillion read/write/select calls, page with size=4 doesnt.
Part of problem seems to be that layout is moving combo boxes popupview
while reflowing. That view is not even shown before you click on combobox,
right? Loading time drops half if you put mShown in line 2265 of
widget/src/gtk/nsWindow.cpp:
< if (mParent && mWindowType == eWindowType_popup ) {
---
> if (mParent && mShown && mWindowType == eWindowType_popup ) {
This brakes popup views (they open at wrong place), but that could be fixed
when thay are shown?
Reporter | ||
Comment 2•25 years ago
|
||
Reporter | ||
Comment 3•25 years ago
|
||
Reporter | ||
Comment 4•25 years ago
|
||
I think rods checkin fixed this. both my testpages renders as fast now.
Marking as fixed.
Status: NEW → RESOLVED
Closed: 25 years ago
Resolution: --- → FIXED
Per Tomi, and the fact that 2954 is fixed. Plus no open issues on slow perf on
Linux with Bugzilla queries recently, marking verified.
Status: RESOLVED → VERIFIED
Updated•9 years ago
|
Product: Core → Core Graveyard
You need to log in
before you can comment on or make changes to this bug.
Description
•