**** BEGIN LOGGING AT Wed Nov 13 02:59:58 2013
Nov 13 09:55:21 <ant_work> lumag: I have to refactor all defconfigs of linux-yocto-tiny-kexecboot: where sculpted for 3.2 and just moved to 3.8 when 3.2 was removed...
Nov 13 09:56:27 <ant_work> lumag: the good news is, shrinked defconfigs are accepted again so the defconfigs will be much shorter (like with 2.6.3x pre-yocto)
Nov 13 09:56:37 <ant_work> s/is/are/
Nov 13 10:00:19 <ant_work> bluelightning: if it is easier, pull from my github fork of meta-hh
Nov 13 10:01:01 <bluelightning> ant_work: ok
Nov 13 10:01:20 <bluelightning> ant_work: it would really help in future if you could use create-pull-request/send-pull-request scripts ;)
Nov 13 10:09:04 <ant_work> you know, it started as a couple of patches, then 3..5...10 ;)
Nov 13 10:12:28 <ant_work> bluelightning: saving you from reading the backlogs, the freaky flash is still unstable but collie fb is repaired
Nov 13 10:14:54 <bluelightning> ant_work: \o/
Nov 13 10:17:27 <ant_work> maybe is just my used collie but I fear we have a CFI driver bug/missing support
Nov 13 10:49:28 <switchgott> Hi to all
Nov 13 10:49:35 <switchgott> I hope you could help me please.
Nov 13 10:49:55 <switchgott> Its about compiling kexec for android x86
Nov 13 10:50:12 <switchgott> does someone know this error from make=
Nov 13 10:50:13 <switchgott> kexec/ifdown.c:40:2: warning: implicit declaration of function 'if_nameindex' [-Wimplicit-function-declaration]
Nov 13 10:50:24 <switchgott> kexec/ifdown.c:47:37: error: increment of pointer to unknown structure
Nov 13 10:53:20 <switchgott> any idea?
Nov 13 10:58:40 <ant_work> I think we have a patch for this
Nov 13 11:02:37 <ant_work> check here  http://cgit.openembedded.org/meta-openembedded/tree/meta-initramfs/recipes-kernel/kexec/kexec-tools-klibc-2.0.2/
Nov 13 11:03:46 <ant_work> switchgott: ah, no, it is for 2.0.4
Nov 13 11:03:51 <ant_work> this one https://github.com/andrea-adami/meta-initramfs/blob/master/recipes-kernel/kexec/kexec-tools-klibc-2.0.4/ifdown_if_nameindex.patch
Nov 13 11:04:35 <ant_work> switchgott: you are compiling against klibc isn't?
Nov 13 11:05:26 <ant_work> unfortunately I've still to finish that 2.0.4 recipe, it fails now the final linking of purgatory
Nov 13 11:37:08 <switchgott> Sorry for my delay
Nov 13 11:37:09 <switchgott> Yes
Nov 13 11:37:21 <switchgott> Id like to get it work on my x86 android tab
Nov 13 11:39:25 <switchgott> Big thanks master
Nov 13 11:39:28 <switchgott> But now this error:
Nov 13 11:39:29 <switchgott> purgatory/arch/i386/vga.c:1:20: fatal error: sys/io.h: No such file or directory
Nov 13 11:40:05 <switchgott> Its inside asm
Nov 13 11:41:31 <switchgott> New error
Nov 13 11:41:31 <switchgott> kexec/arch/i386/x86-linux-setup.o:x86-linux-setup.c:function find_mnt_by_fsname: error: undefined reference to 'setmntent'
Nov 13 11:42:37 <switchgott> Only work not with 2.0.4?But with 2.0.2?
Nov 13 11:48:13 <ant_work> isys.io  https://github.com/andrea-adami/meta-initramfs/blob/master/recipes-kernel/kexec/kexec-tools-klibc-2.0.4/x86_sys_io.patch
Nov 13 11:49:08 <ant_work> 2.0.2 builds ok for many archs like x86, mips, ppc, arm ,...
Nov 13 11:56:31 <switchgott> Big thanks for your help
Nov 13 11:56:36 <switchgott> This error i get with 2.0.4
Nov 13 11:56:37 <switchgott> ifdown.c:(.text+0x7f): undefined reference to `if_nameindex'
Nov 13 11:59:05 <ant_work> yes, there is a patch for that
Nov 13 12:04:17 <switchgott> This one?
Nov 13 12:04:17 <switchgott> https://github.com/andrea-adami/meta-initramfs/blob/master/recipes-kernel/kexec/kexec-tools-klibc-2.0.4/ifdown_if_nameindex.patch
Nov 13 12:04:25 <switchgott> Applyed this but didnt work for me :-(
Nov 13 12:06:02 <ant_work> how can it be...we define it in ifdown.c
Nov 13 12:09:24 <switchgott> Dont know.This is my error
Nov 13 12:09:25 <switchgott> kexec/ifdown.o: In function `ifdown': ifdown.c:(.text+0x7f): undefined reference to `if_nameindex' kexec/arch/i386/x86-linux-setup.o: In function `find_mnt_by_fsname': x86-linux-setup.c:(.text+0xd6a): undefined reference to `setmntent' x86-linux-setup.c:(.text+0xde2): undefined reference to `endmntent' collect2: ld returned 1 exit status make: *** [build/sbin/kexec] Fehler 1
Nov 13 12:11:29 <ant_work> I don't remember the second one...
Nov 13 12:11:42 <ant_work> Try to apply *all* the patches and rebuild
Nov 13 12:11:50 <ant_work> it should gail on purgatory
Nov 13 12:11:53 <ant_work> *fail
Nov 13 12:12:03 <ant_work> fail ist nicht geil ;)
Nov 13 12:13:07 <ant_work> seems purgatory_flags.patch neds more work
Nov 13 12:15:10 <ant_work> hm
Nov 13 12:15:12 <ant_work> fwiw
Nov 13 12:15:18 <switchgott> Do you speek german?
Nov 13 12:15:32 <ant_work> yes but not natively
Nov 13 12:15:37 <ant_work> https://github.com/tias/android-busybox-ndk
Nov 13 12:15:43 <ant_work> mntent -- has patch
Nov 13 12:15:43 <ant_work>     CONFIG_DF -- undefined reference to 'setmntent', 'endmntent'
Nov 13 12:15:50 <ant_work> Androids libc implementation claims to implement the methods in the error, but surprisingly does not.
Nov 13 12:15:55 <ant_work> ah
Nov 13 12:16:07 <switchgott> ??Sorry, verstehe gerade kein Wort :-)
Nov 13 12:16:56 <ant_work> see, seems that like klibc the android libc lack some definitions/something
Nov 13 12:17:12 <switchgott> How can i fix it?
Nov 13 12:17:16 <switchgott> This is my error with 2.0.2
Nov 13 12:17:16 <switchgott> purgatory/../kexec/kexec-sha256.h:5: error: expected specifier-qualifier-list before 'uint64_t' purgatory/purgatory.c:10: error: expected '=', ',', ';', 'asm' or '__attribute__' before 'sha256_digest' purgatory/purgatory.c: In function 'verify_sha256_digest': purgatory/purgatory.c:15: error: 'sha256_digest_t' undeclared (first use in this function)
Nov 13 12:17:29 <switchgott> ./util_lib/include/sha256.h:11: error: expected specifier-qualifier-list before 'uint32_t' ./util_lib/include/sha256.h:16: error: expected '=', ',', ';', 'asm' or '__attribute__' before 'sha256_digest_t' ./util_lib/include/sha256.h:19: warning: type defaults to 'int' in declaration of 'uint8_t' ./util_lib/include/sha256.h:19: error: expected ';', ',' or ')' before '*' token ./util_lib/include/sha256.h:20: error: expected declara
Nov 13 12:17:36 <switchgott> purgatory/purgatory.c:2:20: error: limits.h: No such file or directory purgatory/purgatory.c:3:20: error: stdint.h: No such file or directory
Nov 13 12:18:12 <ant_work> most of these have been fixed for compatibilty with klibc
Nov 13 12:18:23 <ant_work> I have no idea about android work/libc, sorry
Nov 13 12:19:00 <switchgott> uhh
Nov 13 12:19:05 <ant_work> if you can compile it statically with klibc then it is what we did
Nov 13 12:19:12 <switchgott> Do you know a working git for android x86 and kexec?
Nov 13 12:19:44 <ant_work> x86..hmm...don't think so...
Nov 13 12:19:48 <switchgott> and this?
Nov 13 12:19:48 <switchgott> purgatory/purgatory.c:2:24: error: sys/limits.h: No such file or directory
Nov 13 12:20:12 <ant_work> there is achain of errors fixed by the patches I showed you
Nov 13 12:20:33 <ant_work> 2.0.2 builds
Nov 13 12:20:41 <ant_work> 2.0.4 builds up to 99% ;)
Nov 13 12:21:15 <ant_work> but *for klibc* both need the serie of patches we used
Nov 13 12:21:31 <ant_work> for (e)glibc all builds vanilla
Nov 13 12:21:58 <ant_work> again, no idea of android headers
Nov 13 12:22:30 <ant_work> but I'm pretty sure a klibc-statical binary would work
Nov 13 12:25:18 <switchgott> This is my flag for create Makefile.Maybe there is a problem
Nov 13 12:25:24 <switchgott> LDFLAGS=-static ./configure --host=i386 CC=/home/android-ndk-r9/toolchains/x86-4.4.3/prebuilt/linux-x86_64/bin/i686-linux-android-gcc CFLAGS=--sysroot=/home/android-ndk-r9/platforms/android-14/arch-x86/ CPPFLAGS=-I/home/android-ndk-r9/platforms/android-14/arch-x86/ LDFLAGS=--sysroot=/home/android-ndk-r9/platforms/android-14/arch-x86/
Nov 13 12:26:30 <ant_work> I can't compare with mines offhand
Nov 13 12:26:44 <switchgott> Mh :-(
Nov 13 12:26:56 <ant_work> I don't think it is a sysroot problem, just the headers
Nov 13 12:31:37 <switchgott> which header?
Nov 13 12:31:46 <switchgott> Any idea how i can find out?You are my last hope :-((((
Nov 13 12:32:38 <ant_work> see the differences
Nov 13 12:32:39 <ant_work> http://linux.die.net/include/net/if.h
Nov 13 12:32:47 <ant_work> http://git.kernel.org/cgit/libs/klibc/klibc.git/tree/usr/include/net/if.h
Nov 13 12:32:52 <ant_work> http://www.scs.stanford.edu/histar/src/pkg/uclibc/include/net/if.h
Nov 13 12:52:31 <switchgott> Different to which file from me?
Nov 13 12:53:48 <switchgott> With 2.0.2 this error is the problem purgatory/purgatory.c:2:24: error: sys/limits.h: No such file or directory
Nov 13 13:17:59 <switchgott> Damm
Nov 13 13:18:06 <switchgott> With ne compiler it works!Bzt not this
Nov 13 13:18:07 <switchgott> purgatory/purgatory.c:2:28: fatal error: include/limits.h: No such file or directory
Nov 13 13:19:31 <ant_work> this is avoided by one of the patches above (don't remember which one...)
Nov 13 13:19:39 <ant_work> note: for klibc
Nov 13 13:21:19 <ant_work> ah, or this http://www.zytor.com/pipermail/klibc/2012-May/003268.html
Nov 13 13:21:59 <ant_work> sorry but it has been long time...
Nov 13 13:26:17 <switchgott> Uhhh
Nov 13 13:26:17 <switchgott> purgatory/purgatory.c:1:26: fatal error: linux/limits.h: No such file or directory
Nov 13 13:29:55 <switchgott> Why his can find limits.h??
Nov 13 13:35:57 <switchgott> Do you know how i can debug make?
Nov 13 13:38:44 <ant_work> I am sorry, even if both are not standard the android headers are different from the klibc so I'm not sure the same patches will work
Nov 13 13:39:07 <ant_work> you coul dhav ebetter luck asking some #android developer
Nov 13 17:16:06 <ant_home> bluelightning: testing once more the refreshed patchset
Nov 13 17:16:23 <ant_home> I'll send apull-request in a few mins if all is ok
Nov 13 17:16:28 <bluelightning> ant_home: ok, great - thanks
Nov 13 17:20:03 <ant_home> btw, not having h1940 in linux-yocto-dev's COMPATIBLE_MACHINE linux-dummy was built as fall-back
Nov 13 17:20:08 <ant_home> and I got
Nov 13 17:20:11 <ant_home> :WARNING: linux-dummy: No generic license file exists for: GPL in any provider
Nov 13 17:20:18 <ant_home> just FYI
Nov 13 17:20:50 <ant_home> what should one understand?
Nov 13 17:20:53 <ant_home> ;)
Nov 13 17:22:02 <bluelightning> ant_home: I'm sure we'd take a patch to set linux-dummy's LICENSE to GPLv2
Nov 13 17:22:11 <bluelightning> I'm not even sure what linux-dummy is for
Nov 13 17:22:26 <ant_home> yea
Nov 13 18:13:14 <ant_home> ok, sent
Nov 13 19:03:00 <bluelightning> ant_home: so your series contains all recent outstanding patches?
Nov 13 19:05:00 <ant_home> yes
Nov 13 19:08:22 <bluelightning> ok, will merge now
Nov 13 19:09:52 <ant_home> seems h1940 needs some more touch
Nov 13 19:10:44 <bluelightning> so, should I wait? or just merge this set first?
Nov 13 19:10:53 <ant_home> merge pls
Nov 13 19:11:37 <bluelightning> done
Nov 13 19:11:54 <ant_home> I'm building -sato for collie right now, about 50% done
Nov 13 19:12:15 <ant_home> thanks
Nov 13 19:13:28 <ant_home> bluelightning: with the next patches I'll try to slim the kernel down to 1,8.2,5 MB max
Nov 13 19:13:48 <ant_home> then you can retry OPIE
Nov 13 19:13:54 <bluelightning> cool :)
Nov 13 19:14:23 <ant_home> now kernel eats too many resources, maybe on collie ram is enough
Nov 13 19:14:32 <ant_home> poodle is horribly slow in sato
Nov 13 19:14:41 <bluelightning> yeah, I can believe it
Nov 13 19:14:43 <ant_home> (lumag says: add swap ;)
Nov 13 19:14:56 <bluelightning> Sato feels sluggish in qemu on my i5 laptop...
Nov 13 19:15:08 <ant_home> o_O
Nov 13 19:15:22 <ant_home> c7x0 is not so bad...
Nov 13 19:15:35 <bluelightning> I could be wrong but I think it's the way its always been
Nov 13 19:16:04 <ant_home> we have to add that xinput calibration + pointercal in some way
Nov 13 19:16:38 <ant_home> maybe in XSERVER ?
Nov 13 19:17:08 * ant_home feels like having already asked that
Nov 13 19:18:37 <bluelightning> XSERVER would be one way, but I have to admit I'm not sure
Nov 13 19:22:32 <ant_home> bluelightning: heh, now that I've used its kb I see that I have the German model...some letters are strangely mapped on kb
Nov 13 19:23:18 <bluelightning> ah, the beloved QWERTZ
Nov 13 19:23:34 <ant_home> yep
Nov 13 19:41:40 <ant_home> ok, is booting now
Nov 13 19:42:29 <ant_home> hm... strange, it doesn't look right on qvga
Nov 13 19:43:23 <ant_home> and I don't see the pointer
Nov 13 19:50:01 <bluelightning> heading home... bbl
Nov 13 23:00:26 <ant_home> lumag: sigh.. I've read the Status Register as decimal :/
Nov 13 23:01:47 <ant_home> the bitmask is very different... 80h = 128 = 10000000 -> the 7 bits are 0
Nov 13 23:02:34 <ant_home> so the status register is clear
Nov 13 23:03:24 <ant_home> I'm working now on timeouts, buffered write takes >4 cycles on this NOR where on Strataflash is written >2(h3600)
Nov 13 23:25:51 <ant_home> lumag_: hi there
Nov 13 23:25:58 <lumag_> hello ant_home
Nov 13 23:26:48 <ant_home> I have maybe a clue about NOR errors
Nov 13 23:27:19 <ant_home> maybe , I say maybe, there is a missing check for buffer availability
Nov 13 23:27:38 <ant_home> but I don't understand exactly what's going on with mutexes etc..
Nov 13 23:27:59 <ant_home> basically the code reflects the flow-chart of the flash specs
Nov 13 23:28:10 <ant_home> for "write to buffer"
Nov 13 23:28:25 <ant_home> http://lxr.free-electrons.com/source/drivers/mtd/chips/cfi_cmdset_0001.c#L1689
Nov 13 23:29:09 <ant_home> here they start the operation:  CMD(0xe8)
Nov 13 23:30:38 <ant_home> in the examples they check immediately XSR.7
Nov 13 23:30:53 <ant_home> 1=Page Buffer Program
Nov 13 23:30:53 <ant_home> Ready
Nov 13 23:30:53 <ant_home> 0=Page Buffer Program
Nov 13 23:30:53 <ant_home> Busy
Nov 13 23:31:05 <ant_home> what does the code exactly?
Nov 13 23:32:16 <ant_home> (note, the check for the Status Register seems needed as well as in the docs)
Nov 13 23:33:21 <ant_home> but here we talk about the Extended Status Register, accessible only when the the WSM is on
Nov 13 23:33:36 <ant_home> i.e. after 0xe8 command
Nov 13 23:40:40 <ant_home> as far as I see, the code in cfi_cmdset_0001 does that later
Nov 13 23:41:14 <ant_home> 1692                 /* Argh. Not ready for write to buffer */
Nov 13 23:41:14 <ant_home> 1693                 map_word Xstatus = map_read(map, cmd_adr);
Nov 13 23:43:21 <lumag_> Have to check.
Nov 13 23:45:56 <ant_home> btw, I've posted before
Nov 13 23:45:58 <ant_home> <ant_home> lumag: sigh.. I've read the Status Register as decimal :/
Nov 13 23:45:58 <ant_home> <ant_home> the bitmask is very different... 80h = 128 = 10000000 -> the 7 bits are 0
Nov 13 23:45:58 <ant_home> <ant_home> so the status register is clear
Nov 13 23:45:58 <ant_home> <ant_home> I'm working now on timeouts, buffered write takes >4 cycles on this NOR where on Strataflash is written >2(h3600)
Nov 13 23:46:28 <ant_home> I have to add that the Strataflash is slower, not faster
Nov 13 23:46:51 <ant_home> higher timings
Nov 13 23:50:58 <ant_home> so, >sa1100: Chip not ready for buffer write. Xstatus = 80808080, status = 80808080
Nov 13 23:51:41 <ant_home> 1700                 printk(KERN_ERR "%s: Chip not ready for buffer write. Xstatus = %lx, status = %lx\n",
Nov 13 23:51:59 <ant_home> this happens afterwards
Nov 13 23:52:10 <ant_home> like you should have checked before...
Nov 14 00:01:45 <ant_home> bluelightning: lumag_: one of you should test the flash: one run is enough if it fails there as well...
Nov 14 00:01:55 <ant_home> have to go now
Nov 14 00:01:56 <ant_home> gn
Nov 14 00:02:00 <lumag_> gn
**** ENDING LOGGING AT Thu Nov 14 02:59:59 2013