Closed Bug 459290 Opened 16 years ago Closed 14 years ago

Canvas view and rest of UI not responsive while loading one image

Categories

(Firefox for Android Graveyard :: General, defect)

x86
Linux
defect
Not set
normal

Tracking

(Not tracked)

RESOLVED WORKSFORME

People

(Reporter: romaxa, Unassigned)

References

()

Details

Attachments

(3 files)

Preconditions: add URL to the bookmarks list Steps: 1) Open fennec with URL 'file:///' See page with content of root file system... Page is scrollable and responsive... 2) Open bookmarks and select bookmark with URL. while it is loading image content, try to scroll, zoom, pan, open tab/control bars ... EXPECTED: canvas is pannable, UI responsive, user happy ACTUAL: canvas, UI freeze while fennec loading image content... and not respond to user during long time.
Blocks: 441364
Attached file Oprofile data (deleted) —
Attached file random break while image loading (deleted) —
Here's some other oprofile data, with a few patches applied. Loading the image took a total of 5 seconds, including the final scale/draw; the initial part of the image was displayed within < 1s. There's something weird going on inside the kernel, the lzo1x_decompress_safe is usually a result of a jffs2 read but I have no idea what would be being read... 36225 52.2584 vmlinux-oprofile-diablo-200835 16997 24.5200 libxul.so 4068 5.8685 Xomap 1933 2.7886 libmozjs.so 1709 2.4654 oprofiled 1509 2.1769 libc-2.5.so 746 1.0762 libglib-2.0.so.0.1200.12 samples % app name symbol name 7226 10.5090 vmlinux-oprofile-diablo-200835 lzo1x_decompress_safe 4747 6.9037 libxul.so ycc_rgb_convert_argb 3252 4.7295 vmlinux-oprofile-diablo-200835 memcpy 2290 3.3304 vmlinux-oprofile-diablo-200835 shrink_active_list 2226 3.2373 vmlinux-oprofile-diablo-200835 crc32_le 1977 2.8752 libxul.so jpeg_idct_islow 1709 2.4855 oprofiled /usr/bin/oprofiled 1534 2.2309 Xomap fbSolid 1509 2.1946 libc-2.5.so /lib/libc-2.5.so 1337 1.9444 Xomap fbmemcpy_arm 1147 1.6681 libxul.so h2v2_fancy_upsample 1097 1.5954 vmlinux-oprofile-diablo-200835 omap2_pm_idle 860 1.2507 vmlinux-oprofile-diablo-200835 __copy_to_user 788 1.1460 vmlinux-oprofile-diablo-200835 __copy_from_user 760 1.1053 vmlinux-oprofile-diablo-200835 clk_enable 746 1.0849 libglib-2.0.so.0.1200.12 /usr/lib/libglib-2.0.so.0.1200.12 vmlinux: 7226 19.9476 lzo1x_decompress_safe 3252 8.9772 memcpy 2290 6.3216 shrink_active_list 2226 6.1449 crc32_le 1097 3.0283 omap2_pm_idle 860 2.3741 __copy_to_user 788 2.1753 __copy_from_user 760 2.0980 clk_enable 634 1.7502 __memzero 625 1.7253 flush_pfn_alias 569 1.5707 omap2_onenand_wait 505 1.3941 inflate_fast 443 1.2229 page_check_address 406 1.1208 __do_softirq 394 1.0876 schedule libxul: samples % app name symbol name 4747 25.1830 libxul.so ycc_rgb_convert_argb 1977 10.4881 libxul.so jpeg_idct_islow 1147 6.0849 libxul.so h2v2_fancy_upsample 570 3.0239 libxul.so decode_mcu 239 1.2679 libmozjs.so JS_CallTracer 225 1.1936 libmozjs.so js_TraceObject
xulrunner - 20081010-03:29-20239:bd8d176ed481 fennec - 20081010-01:51-235:3772037edcd8 In my case it "thinking" ~10 sec before displaying image and ~8 sec after displaying image... With the same xulrunner build but TestGtkEmbed it immediately show first picture parts, and show progress of picture loading... it takes ~3-4 seconds I think something wrong with fennec not with kernel...
Attached file TestGtkEmbed profile (deleted) —
vlad, kernel samples are largly irrelevant. Probably you just captured some background noise caused by syslogd flushing something or stuff, jffs2 is largely inpredictible. We should concentrate on userspace, my gut feeling kernel has nothing to do here.
WFM in Fennec trunk.
Status: NEW → RESOLVED
Closed: 14 years ago
Resolution: --- → WORKSFORME
You need to log in before you can comment on or make changes to this bug.

Attachment

General

Created:
Updated:
Size: