**** BEGIN LOGGING AT Tue May 22 03:00:04 2018 May 22 08:21:11 ant_work: I am just testing a new kexecboot kernel (ext4 issue) May 22 08:21:41 when is kexecboot 0.7 ready ? I just got it from git May 22 08:22:36 also I noticed newer "kexec" need to pass --atags parameter otherwise "Could not scan "/proc/device-tree/": No such file or directory" May 22 08:22:56 hi greguu May 22 08:23:05 hi ant_work May 22 08:23:48 I did not test toroughly last kexecboot May 22 08:23:54 it appears solid May 22 08:24:20 but as you see there are new issues May 22 08:24:26 on Zaurus atags is ignored May 22 08:26:24 when I test kexecboot on ArchLinux the kexec requires --atags May 22 08:26:32 must be a different version May 22 08:26:45 what are the new issues ? did I miss something May 22 08:30:07 well, behind the corner is gcc8 / armv5te only May 22 08:30:25 I am upgrading collie in OE now that we still have gcc7 May 22 08:31:02 I can't port the 4.2 patches to 4.17...not now May 22 08:31:07 ah, will xscale and iwmmxt be dropped ? May 22 08:31:39 hm, I've read armv5 will be dropped, armv5te the new minimum May 22 08:32:00 must check about extensions May 22 08:32:09 armv5te should be fine for Cxx00 May 22 08:32:16 but we can still use gcc7 now May 22 08:32:31 have you tried 4.17-rc? May 22 08:32:45 I am tempted to commit 4.14, last working one May 22 08:33:01 no I am still with 4.14.x May 22 08:33:11 I played with 4.16 but not for long May 22 08:33:29 so, 4.16 is full of issues you said May 22 08:34:35 ok then, I have 4.16.x ready but not tested on real hw. May 22 08:34:53 I'll commit 4.14 in OE then start 4.17 on LinuxPDA github May 22 08:35:14 I did not test much, pcmcia issue was still there and it was behaving odd in suspend / resume May 22 08:35:27 as for kexecboot, let me ask Tony if he wants that we add the new --no-checks option May 22 08:35:44 this would be more usefull for Android right ? May 22 08:35:45 I'd do that, ask Jay7 then release May 22 08:35:48 yes May 22 08:35:58 we will need it for the gemini May 22 08:36:03 most issues coming from purgatory code May 22 08:36:04 I should get my unit next month May 22 08:36:17 great May 22 08:39:10 greguu, about atags, the kernel lost the ability of reading Zaurus atags long ago, with 2.6 May 22 08:39:39 there is an option about legacy atags in kernel but it does not apply May 22 08:39:54 zaurus bootloader had the old-style atags May 22 08:40:11 I can't see how kexec could parse it May 22 08:40:20 (without extra code) May 22 08:41:26 kexec-tools 2.0.15 on archlinux, when testing kexecboot, I get "unrecoverable error: could not scan "/proc/device-tree/": as it tries to use the host os kexec-utils May 22 08:41:46 it tries to use dtb May 22 08:42:00 this also will not work May 22 08:45:36 I am looking at the patch which added dtb to kexecboot, it does not do anything suspicious lik ethat... May 22 08:46:00 it think its a kexec-utils thing May 22 08:46:02 and I am sure I tested that one and a few commits more May 22 08:46:33 I am working on 2.017+git sources May 22 08:46:36 kexec in my initramfs is 2.0.2 May 22 08:46:54 the old kexec-klibc-static was 2.0.2 May 22 08:47:05 I am still using that one May 22 08:47:19 and it works, just testing on arch fails with 2.0.15 May 22 08:49:21 I can upload the precompiled new one May 22 08:49:33 I did crossbuild for 7 archs ;) May 22 08:50:35 ok, I can test it May 22 09:04:38 I'll send you first by email in a few mins May 22 09:09:06 got news about my gemini... "Your parcel has left the United Kingdom and is on its way"... finally :) May 22 09:12:00 hi May 22 09:12:11 that's great to hear omegamoon May 22 09:12:18 mine is still MIA May 22 09:13:23 but at least yesterday I received my Gemini t-shirt from the Facebook competition :) It's great though it's too big for me, I have chosen the wrong size May 22 10:09:25 mmount May 22 10:17:08 greguu, I've sent it now, for spitz May 22 10:42:06 ant_work: thanks, I will give it a go, but I want to fix the ext4 issue first May 22 11:05:46 how did you create the filesystem? May 22 11:06:08 on Arch, with a card-reader? May 22 11:06:33 on the Zaurus, on Arch on other Linux May 22 11:06:39 mkfs.ext4 /dev/mmcblk0p1 May 22 11:06:55 it may work if mkfs.ext4 /dev/mmcblk0p1 -O^metadata_csum May 22 11:06:59 see, OpenEmbedded is doing right now the upgrade 1.43.8 -1.44.2 ... May 22 11:07:08 I've sent a warning May 22 11:07:58 Tony said mkfs.ext4 -O ^64bit might work on newer kernels May 22 11:09:28 greguu, https://ext4.wiki.kernel.org/index.php/Ext4_Disk_Layout May 22 11:09:36 CHECK AFTER 0X52 May 22 11:09:48 These fields are for EXT4_DYNAMIC_REV superblocks only. May 22 11:09:48 Note: the difference between the compatible feature set and the incompatible feature set is that if there is a bit set in the incompatible feature set that the kernel doesn't know about, it should refuse to mount the filesystem. May 22 11:09:48 e2fsck's requirements are more strict; if it doesn't know about a feature in either the compatible or incompatible feature set, it must abort and not try to meddle with things it doesn't understand... May 22 11:10:52 well, the fstypes in kexecboot understand only ext3 superblocks May 22 11:11:22 but afaik it should still mount it read only if aware May 22 11:11:58 hmm I have kexecbooted many ext4 over the years May 22 11:12:07 yes we all have May 22 11:12:11 it's just if you create it right now May 22 11:12:27 INCOMPAT_CSUM_SEED May 22 11:13:46 ahm, I thought only INCOMPAT_LARGEDIR May 22 11:14:05 no, I thinkg is the checksum thing May 22 11:14:48 and INCOMPAT_EA_INODE -> stores large extended attributes in an external inode May 22 11:15:17 greguu, yep May 22 11:15:18 The metadata_csum feature (RO_COMPAT_METADATA_CSUM) is now enabled by default by mke2fs. May 22 11:16:12 but it should still mount read only May 22 11:18:03 I think its the kernel at fault not kexecboot May 22 11:20:18 right, mount just calls an ioctl so it is kernel here May 22 11:20:43 the FS is detected as ext4 May 22 11:21:58 yes but the kernel finds an unknown incompatible feat and refuses to mount wiyth unknown option May 22 11:23:06 I looked at ext4.h in 3.10.y and all these feartures are there May 22 11:25:29 see, official docs should be: May 22 11:25:31 https://www.kernel.org/doc/Documentation/filesystems/ext4.txt May 22 11:25:36 https://ext4.wiki.kernel.org/index.php/Ext4_Howto May 22 11:26:07 ext4 has a long and troubled, history, though May 22 11:26:50 looks like these guys came across the same issue: https://github.com/lgeek/okreader/issues/17 May 22 11:28:45 I think key is the version of ioctl EXT4_IOC_GETFLAGS May 22 11:30:48 ..and ofc the changes your distro does to mke2fs.conf May 22 11:31:09 we can tell people to format with mkfs.ext4 /dev/mmcblk0p1 -O^metadata_csum May 22 11:31:21 or use f2fs May 22 11:32:46 3.10.y ioctl.c has RO_COMPAT_METADATA_CSUM checks May 22 11:33:08 so 3.10.y should be fine to drive this newer ext4 May 22 11:33:41 can we debug / verbose the mount output in kexecboot ? May 22 11:34:47 I see there is a 'verbose' option May 22 11:34:57 in klibc do_mounts.c May 22 11:35:40 so when we first mopunt ro we could add May 22 11:35:57 'ro verbos' instead May 22 11:36:10 but no, atm is not implemented May 22 11:40:13 what kernel version did you use last for a kexecboot kernel ? May 22 11:41:39 I will play with it again tomorrow.. May 22 11:41:42 * greguu is gone for the day May 22 11:42:11 https://github.com/LinuxPDA/kexecboot/blob/master/src/kexecboot.c#L344 May 22 11:42:14 ^ May 22 11:42:34 (sorry today phone calls no-stop) May 22 11:43:13 you can add verbose here May 22 11:44:02 last ver was 4.14 May 22 11:44:16 but uploaded only 4.4 May 22 11:44:28 EXT4_IOC_GETFLAGS May 22 11:44:33 argh May 22 11:44:35 https://github.com/LinuxPDA/linux-kexecboot/tree/master/zaurus May 22 11:46:39 {"verbose", MS_VERBOSE, MS_VERBOSE, 0}, May 22 11:46:59 {"ro", MS_RDONLY, MS_RDONLY, 0}, May 22 11:47:42 just add here in https://github.com/LinuxPDA/kexecboot/blob/master/src/kexecboot.c#L344 **** ENDING LOGGING AT Wed May 23 03:00:01 2018