IRC log of #maemo-ssu for Sunday, 2012-12-30

« Saturday, 2012-12-29 Index Monday, 2012-12-31 »
*** tg has quit IRC00:17
*** tg has joined #maemo-ssu00:34
DocScrutinizer05~dict plow00:36
infobotDictionary 'plow' (1 of 8): \Plow\, Plough \Plough\ (plou), v. i. To labor with, or as with, a plow; to till or turn up the soil with a plow; to prepare the soil or bed for anything. --Shak. [1913 Webster] Doth the plowman plow all day to sow ?   --Isa. xxviii. 24. [1913 Webster] Plowable.00:36
*** andre__ has quit IRC00:59
*** BCMM has joined #maemo-ssu01:32
*** joshgillies has joined #maemo-ssu03:06
*** joshgillies has quit IRC03:15
*** andre__ has joined #maemo-ssu03:16
*** andre__ has joined #maemo-ssu03:16
*** M4rtinK has quit IRC04:01
*** NIN101 has quit IRC04:02
*** kolp has quit IRC04:08
*** arcean has quit IRC04:09
*** AndrewX192 has quit IRC04:34
*** AndrewX192 has joined #maemo-ssu04:34
*** AndrewX192 has joined #maemo-ssu04:34
*** AndrewX192 has quit IRC04:41
*** AndrewX192 has joined #maemo-ssu04:41
*** AndrewX192 has joined #maemo-ssu04:41
*** AndrewX192_ has joined #maemo-ssu04:46
*** AndrewX192_ has joined #maemo-ssu04:46
*** AndrewX192 has quit IRC04:47
*** AndrewX192_ has quit IRC04:48
*** AndrewX192 has joined #maemo-ssu04:51
*** BCMM has quit IRC04:52
*** AndrewX192 has quit IRC04:56
*** AndrewX192 has joined #maemo-ssu04:57
*** AndrewX192 has joined #maemo-ssu04:57
*** AndrewX192 has quit IRC04:58
*** AndrewX192 has joined #maemo-ssu05:02
*** AndrewX192 has joined #maemo-ssu05:07
*** AndrewX192 has joined #maemo-ssu05:07
*** AndrewX192_ has joined #maemo-ssu05:30
*** AndrewX192_ has joined #maemo-ssu05:30
*** AndrewX192_ has quit IRC05:32
*** AndrewX192 has quit IRC05:34
*** AndrewX192 has joined #maemo-ssu05:35
*** amiconn has quit IRC05:53
*** amiconn_ has joined #maemo-ssu05:53
*** amiconn_ is now known as amiconn05:53
jon_ymerlin1991: ping05:56
*** DocScrutinizer05 has quit IRC06:03
*** DocScrutinizer05 has joined #maemo-ssu06:03
jon_ymerlin1991: did you ever encounter something about a missing libfakeroot-sysv.so when doing http://wiki.maemo.org/CSSU-thumb_toolchain_setup_%28gcc4.7.2-linaro%29 ?06:03
*** nox- has quit IRC06:10
merlin1991nope, but I never set one up06:14
merlin1991freemangordon: ^^06:14
*** Martix has quit IRC06:27
*** _xnt14 has joined #maemo-ssu06:36
jon_ymerlin1991: I copied the so from FREMANTLE_ARMEL06:44
jon_yno idea if it is OK06:44
jon_ymerlin1991: also, is there a way for scratchbox to access outside the chroot?06:47
merlin1991nope07:18
*** dhbiker has joined #maemo-ssu08:56
jon_yoh haha wow, I've been building the kernel wrong all these time09:21
jon_yneed to extract kernel sources first into the git checkout09:21
jon_ynothing about that in the README09:21
*** FIQ has quit IRC10:04
*** jonwil has joined #maemo-ssu10:29
*** Estel_ has joined #maemo-ssu11:12
*** Estel_ has quit IRC11:12
*** Estel_ has joined #maemo-ssu11:12
*** Estel_ is now known as edchngtf11:15
*** edchngtf is now known as Estel_11:15
*** Estel_ has quit IRC11:16
*** M4rtinK has joined #maemo-ssu12:29
*** freemangordon has quit IRC12:29
*** NIN101 has joined #maemo-ssu13:03
*** BCMM has joined #maemo-ssu14:05
*** arcean has joined #maemo-ssu14:21
*** freemangordon has joined #maemo-ssu14:28
freemangordonmerlin1991: x8614:30
*** kolp has joined #maemo-ssu14:41
*** luf has joined #maemo-ssu14:46
luffreemangordon, merlin1991: what about thle libxml2 merge request? I think we agreed it's bug and should be fixed.14:59
lufhttps://gitorious.org/community-ssu/libxml2/merge_requests/114:59
DocScrutinizer05((is there a way for scratchbox to access outside the chroot?)) depends on definition of "access". You can do all sorts of wget apt-get etc. You also can bindmount dirs form host into chroot, that's how a chroot works15:03
DocScrutinizer05actually SB can do apt-get only because host's /dev is bindmounted into the chroot15:04
DocScrutinizer05those bindmounts are also the reason why it's a VERY BAD idea to rm -rf /scratchbox15:05
DocScrutinizer05ask mohammad if you wanna know how bad an idea that is15:05
jon_yDocScrutinizer05: I forgot about bind mount15:20
lufgregoa: ping15:30
gregoaluf: You sent me a contentless ping.  This is a contentless pong.  Please provide a bit of information about what you want and I will respond when I am around.15:30
lufI built libxml2 deb package and it put into depends (using shlibs) zlib1g (>= 1:1.2.1). But I patch it and the required version of zlib1g should be 1.2.3. I don't see where the shlibs takes the 1.2.1 version.15:32
lufgregoa: ^^^15:32
*** arcean_ has joined #maemo-ssu16:17
*** arcean_ has quit IRC16:18
*** arcean_ has joined #maemo-ssu16:18
*** zogg_ has joined #maemo-ssu16:19
*** arcean has quit IRC16:21
*** zogg__ has quit IRC16:21
*** arcean has joined #maemo-ssu16:23
*** arcean_ has quit IRC16:25
*** Martix has joined #maemo-ssu16:53
gregoaluf: zlib1g has a shlibs file that says: "libz 1 zlib1g (>= 1:1.2.1)" and that gets inserted in your libxml2 builg via the "${shlib:Depends}" in debian/control. if you need 1.2.3, then just adding "zlib1g (>= 1:1.2.3)" to the Depends line in debian/control should be enough. (or bumping the shlibs in zlib source package and rebuild and build libxml2 against that one)17:13
*** Martix has quit IRC17:17
*** arcean_ has joined #maemo-ssu17:30
freemangordonluf: https://gitorious.org/community-ssu/libpng17:38
*** arcean_ has quit IRC17:38
luffreemangordon: coool17:53
*** jonwil has quit IRC17:53
luffreemangordon: We need to revert commits with the NEON opts I added.17:54
freemangordonluf: I am testing now with test suite on png website, so far so good17:54
freemangordonluf: no, my changes are on top of those17:54
freemangordonluf: BTW I don;t guarantee there won;t be some buffer overflow, please check aligment part :D17:55
lufAhhh I see.17:55
freemangordonluf: well, it should be ok, but i'll prefer another one to review it17:56
freemangordon*i'd17:56
luffreemangordon, merlin1991: what about thle libxml2 merge request? I think we agreed it's bug and should be fixed.17:59
freemangordonI am ok18:00
freemangordonbut only merlin1991 can merge18:00
lufok.18:00
freemangordonluf: BTW libpng neon patch can be used without alignment part, but performance is worse18:02
freemangordonnow it scores 2.220, unaligned ~2.40018:03
lufNice.18:03
luffreemangordon: png_read_filter_row: unsigned int bpp = (png_ptr->pixel_depth + 7) >> 3; - what does it do?18:14
freemangordoncalculates bytes per pixel18:14
freemangordonbut htis is original code18:14
lufI see.18:14
luffreemangordon: did you add prev_row to the struct?18:16
freemangordonno18:16
freemangordonI didn't touch ABI18:16
lufI see stupid question ...18:16
luffreemangordon: just one thing. 15 upstream has png_ptr->old_prev_row_size = row_bytes + 64; below row_buf setting. I see you keep it above ...18:21
lufs/old_prev_row_size/old_big_row_buf_size/18:22
infobotluf meant: freemangordon: just one thing. 15 upstream has png_ptr->old_big_row_buf_size = row_bytes + 64; below row_buf setting. I see you keep it above ...18:22
freemangordonluf: lemme check18:28
lufAlso I don't think it's good to set old_prev_row_size prior to prev_row allocation.18:29
lufI think also prev_row should be 16-bit aligned.18:29
lufBased on libpng15 code: png_ptr->row_buf = temp - extra - 1/*filter byte*/; png_ptr->prev_row = temp - extra - 1/*filter byte*/;18:30
freemangordonluf: your first wuestio: it does not really matter which one will be set first18:30
freemangordon*question18:30
freemangordonluf: prev_row alignment is done in real-time, that was the hack I was talking aboot18:31
freemangordonupstream 15 has an additional struct member to keep unaligned pointer18:31
freemangordonwe don't :)18:31
freemangordonso I keep unaligned pointer and calculate/pass aligned when needed18:32
freemangordona bit of overhead, but no option there aiui18:32
freemangordonnot that I like that much :D18:33
lufSo row_bytes is always 16-bit aligned?18:33
freemangordonno18:33
freemangordonaiui we alloc row_bytes+6418:33
freemangordonthat's called big_row_buf or smtg18:34
freemangordonrow pointer is aligned within this buffer18:34
freemangordonso we have padding before and after the real row18:35
freemangordonI am not sure i fully understand why this padding is needed, that's why i wanted a code review18:35
*** dhbiker has quit IRC18:36
freemangordonvalgrind reports no bad memory access, no memleaks :)18:38
*** dhbiker has joined #maemo-ssu18:38
lufwhat does png_memset_check do?18:41
freemangordonNFC, but I gues it does memset :D18:43
lufSo what's the reason to align16 in png_mem_check?18:45
lufI think it's  a bug.18:45
freemangordonluf: which line?18:45
luf317918:46
freemangordonluf: png_memset_check(png_ptr, align16(png_ptr->prev_row), 0,18:46
freemangordonpng_ptr->rowbytes + 1);18:46
freemangordonluf: that code?18:46
lufyes.18:47
freemangordonno, it is ok, we don't need the whole buffer zeroed18:47
freemangordononly the part that is used18:47
lufYeah allocated is rowbytes+64 so it doesn't overflow.18:48
freemangordonpng_ptr->rowbytes is the "real" row bytesize, not the "big buffer" size18:48
lufI don't see why the buffer around is needed. But I see no problem in you code.18:50
lufJust one more question. Why you changed if ((png_uint_32)row_bytes + 1 > (png_uint_32)65536L) ?18:50
lufI see no reason for such change.18:51
freemangordonluf: aiui I think the padding is just in case :D18:57
freemangordonluf: which line is that change?18:58
freemangordonluf: BTW 64 bytes is a bit of overkill, upstream 1.5 use 4818:59
freemangordonluf: re why i changed ^^^ - aiui this code becomes redundant, TBH I don;t see a reason for it to be there even in not-patched version19:01
freemangordon*non-patched19:01
freemangordonluf: results for http://en.wikipedia.org/wiki/File:PNG_transparency_demonstration_2.png - http://pastebin.com/UYvdgcsb19:09
freemangordonimpressive19:09
kerioheh19:11
keriodo the test on the true transparent one19:11
freemangordonkerio: wanna do it yourself? i'll upload the binaries19:16
keriois the thumby deb for libpng somewhere?19:16
freemangordonjust a minute19:17
freemangordonkerio: http://merlin1991.at/~freemangordon/libpng/19:18
freemangordon.deb is thumb-compiled, despite the name19:19
freemangordonkerio: transparency shhould not affect decoding time19:22
luffreemangordon: You're right. That code has to write some drunken dev. As "Row has too many bytes to allocate in memory." is behind some memory (re)allocation :D19:23
luffreemangordon: Frommy POV the align part is ok.19:24
lufmerlin1991: ping19:26
luffreemangordon: more interesting should be ARM againist NEON with same thumb/non-thumb and gcc.19:28
lufHowever 30% speed up is nice ;)19:28
keriofreemangordon: it's ok if pngtest.png and pngout.png are different, right?19:29
lufkerio: it should be ... due to zlib etc.19:29
lufkerio: let's try rename the pngout.png and test it on that new ;)19:31
freemangordonluf: what do you mean ARM against NEON?19:32
lufhttp://pastebin.com/UYvdgcsb19:32
lufI see ARM, gcc 4.2.1 vs NEON+THUMB, gcc 4.7.2:19:33
kerioCPU time used = 31.420 seconds (decoding 2.370, encoding 29.040 , other 0.010 seconds)19:33
freemangordonluf: THUMB makes almost no difference19:34
keriohold on, i already had a thumby one19:34
freemangordonsee encoding times19:34
freemangordonkerio: yes19:34
luffreemangordon: and different gcc also?19:34
freemangordonlooks like19:34
lufok.19:34
kerioCPU time used = 30.800 seconds (decoding 1.580, encoding 29.170 , other 0.050 seconds)19:35
kerioholy balls19:35
keriofreemangordon: what's this new libpng?19:35
kerioNEONized one?19:35
freemangordonluf: why don't you test it, pngtest binary is here  http://merlin1991.at/~freemangordon/libpng/19:35
freemangordonkerio: yes19:35
lufkerio: yes NEONized + memory alignment19:35
keriothumbify it, too!19:35
freemangordonit is thumbified19:36
*** _xnt14 has quit IRC19:36
luffreemangordon: I'm still running non-thumb version.19:36
freemangordonluf just build the .deb dammit :P19:36
freemangordonluf: ok, i'll build it for you19:36
freemangordonhold on a minute19:37
lufBTW I see several reboots on my dev N900 yesterday. I installed tklocks and others from cssu-devel.19:37
luffreemangordon: don't do it.19:37
lufI'm leaving now ...19:37
freemangordonluf: reboots?19:37
freemangordondo you have sp-rich-core installed?19:37
freemangordonluf: before you leave19:38
luffreemangordon: I have to leave now. I'll continue later or tomorrow.19:38
freemangordonwhere is neon-patched zlib?19:38
freemangordonok19:38
lufNo I don't have.19:39
freemangordonok19:39
lufYou know where is the source.19:39
lufhttp://merlin1991.at/~luf19:39
lufhttp://merlin1991.at/~luf/zlib/source/19:39
lufBTW today's benchmarks showed me less improvement zlib ...19:40
*** luf has quit IRC19:41
*** arcean has quit IRC19:43
*** dafox has joined #maemo-ssu19:56
keriofreemangordon: btw, /dev/shm is neat to do tests19:57
*** MrPingu has joined #maemo-ssu20:11
*** Martix has joined #maemo-ssu20:52
*** Martix has quit IRC20:57
*** Martix has joined #maemo-ssu20:58
*** arcean has joined #maemo-ssu21:28
*** arcean has quit IRC22:08
*** arcean has joined #maemo-ssu22:08
*** MrPingu has quit IRC22:19
*** andre__ has quit IRC22:19
*** Skry has quit IRC22:19
*** MrPingu has joined #maemo-ssu22:19
*** andre__ has joined #maemo-ssu22:19
*** Skry has joined #maemo-ssu22:25
*** Skry has quit IRC22:37
*** Skry has joined #maemo-ssu22:38
*** dhbiker has quit IRC23:09
*** joshgillies has joined #maemo-ssu23:37
*** Oposum has joined #maemo-ssu23:59
« Saturday, 2012-12-29 Index Monday, 2012-12-31 »

Generated by irclog2html.py 2.15.1 by Marius Gedminas - find it at mg.pov.lt!