[09:57] <ppisati> ro
[09:57] <ppisati>  / ro
[09:57] <ppisati> [ 6827.767493] ata1.00: exception Emask 0x0 SAct 0xffffff SErr 0x880000 action 0x6 frozen
[09:58] <ppisati> [ 6827.767497] ata1: SError: { 10B8B LinkSeq }
[09:58] <ppisati> [ 6827.767499] ata1.00: failed command: WRITE FPDMA QUEUED
[09:58] <ppisati> [ 6827.767516] ata1.00: cmd 61/08:00:90:4f:4a/00:00:06:00:00/40 tag 0 ncq 4096 out
[09:58] <ppisati> [ 6827.767516]          res 40/00:00:00:4f:c2/00:00:00:00:00/00 Emask 0x4 (timeout)
[09:58] <ppisati> [ 6827.767517] ata1.00: status: { DRDY }
[09:58] <ppisati> [ 6827.767518] ata1.00: failed command: WRITE FPDMA QUEUED
[09:58] <ppisati> [ 6827.767521] ata1.00: cmd 61/68:08:c0:b9:47/00:00:0d:00:00/40 tag 1 ncq 53248 out
[09:58] <ppisati> [ 6827.767521]          res 40/00:00:00:00:00/00:00:00:00:00/00 Emask 0x4 (timeout)
[09:58] <ppisati> [ 6827.767522] ata1.00: status: { DRDY }
[09:58] <ppisati> ...
[09:58] <ppisati> [ 6828.122829] ata1.00: device reported invalid CHS sector 0
[09:58] <ppisati> [ 6828.122836] sd 0:0:0:0: [sda]
[09:58] <ppisati> [ 6828.122837] Result: hostbyte=DID_OK driverbyte=DRIVER_SENSE
[09:58] <ppisati> [ 6828.122838] sd 0:0:0:0: [sda]
[09:58] <ppisati> [ 6828.122839] Sense Key : Aborted Command [current] [descriptor]
[09:58] <ppisati> [ 6828.122841] Descriptor sense data with sense descriptors (in hex):
[09:58] <ppisati> [ 6828.122842]         72 0b 00 00 00 00 00 0c 00 0a 80 00 00 00 00 00
[09:58] <ppisati> [ 6828.122846]         00 00 00 00
[09:58] <ppisati> [ 6828.122848] sd 0:0:0:0: [sda]
[09:58] <ppisati> [ 6828.122850] Add. Sense: No additional sense information
[09:58] <ppisati> [ 6828.122851] sd 0:0:0:0: [sda] CDB:
[09:58] <ppisati> [ 6828.122852] Write(10): 2a 00 06 4a 4f 90 00 00 08 00
[09:58] <ppisati> [ 6828.122856] end_request: I/O error, dev sda, sector 105533328
[09:58] <ppisati> [ 6828.122858] quiet_error: 3 callbacks suppressed
[09:59] <ppisati> [ 6828.122859] Buffer I/O error on device sda1, logical block 13191410
[09:59] <ppisati> [ 6828.122860] lost page write due to I/O error on sda1
[09:59] <ppisati> ...
[09:59] <ppisati> [ 6828.123406] Buffer I/O error on device sda1, logical block 9537386
[09:59] <ppisati> [ 6828.123407] Buffer I/O error on device sda1, logical block 9537387
[09:59] <ppisati> [ 6828.123409] EXT4-fs warning (device sda1): ext4_end_bio:250: I/O error writing to inode 3812130 (offset 995328 size 8192 starting block 9537644)
[09:59] <ppisati> [ 6828.123411] ata1: EH complete
[09:59] <ppisati> [ 6828.123655] EXT4-fs (sda1): Remounting filesystem read-only
[09:59] <ppisati> [ 6828.123658] EXT4-fs error (device sda1) in ext4_free_blocks:4700: Journal has aborted
[09:59] <ohsix> eh
[09:59] <ppisati> [ 6828.123667] journal commit I/O error
[09:59] <ppisati> [ 6828.123697] EXT4-fs error (device sda1) in ext4_dirty_inode:4610: Journal has aborted
[09:59] <ppisati> [ 6828.123740] EXT4-fs (sda1): ext4_da_writepages: jbd2_start: 9223372036854775807 pages, ino 9213275; err -30
[09:59] <ppisati> [ 6828.123768] EXT4-fs error (device sda1): ext4_journal_start_sb:370: Detected aborted journal
[09:59] <ppisati> [ 6828.123803] EXT4-fs (sda1): ext4_da_writepages: jbd2_start: 9223372036854775807 pages, ino 9213273; err -30
[09:59] <ppisati> [ 6828.123844] EXT4-fs error (device sda1) in ext4_orphan_add:2420: Journal has aborted
[09:59] <ppisati> [ 6828.124019] EXT4-fs error (device sda1) in ext4_orphan_add:2420: Journal has aborted
[09:59] <ppisati> [ 6828.124021] EXT4-fs error (device sda1) in ext4_orphan_add:2420: Journal has aborted
[09:59] <ppisati> [ 6828.124083] EXT4-fs error (device sda1) in ext4_reserve_inode_write:4483: Journal has aborted
[09:59] <ppisati> [ 6828.124158] EXT4-fs error (device sda1): ext4_journal_start_sb:370: Detected aborted journal
[09:59] <ppisati> [ 6828.124258] EXT4-fs error (device sda1) in ext4_reserve_inode_write:4483: Journal has aborted
[09:59] <ppisati> ok, need to reboot and fsck
[10:02] <cking> ppisati, bitten again?
[10:02] <ppisati> yes
[10:03] <ppisati> haswell and intel ssd
[10:03] <cking> ppisati, what kind of workload?
[10:03] <ppisati> i need to find if there's a new firmware for that ssd now
[10:03] <cking> which SSD?
[10:03] <ohsix> ppisati: smart logs can tell you where and why those WRITE FPDMA QUEUED errors happened
[10:03] <ppisati> cking: nothing
[10:03] <ppisati> cking: editing files, terminal with mutt, irssi and some sshs
[10:03] <ppisati> cking: chrome was open too
[10:04] <ppisati> ohsix: but i guess i need smartd running, right?
[10:04] <cking> that's what I saw on my  Intel 520
[10:04] <smb> ppisati, If I were around I would first suggest to see whether the "Buffer I/O errors" are within the drive/partition space
[10:05] <cking> smb, you on vacation?
[10:05] <ohsix> ppisati: nope, smartctl -l error /dev/device
[10:05] <smb> cking, Its "no-work-friday"
[10:05] <cking> smb, you're as bad as me
[10:05] <ohsix> the device stores a bunch of logs, an error log is one of them
[10:06] <ppisati> ohsix: SMART Error Log not supported
[10:06] <cking> ohsix, not sure if it is supported on some intel SSDs
[10:06] <cking> oh, that's what I get too
[10:07] <ppisati> smb: it mentioned sda1, so it should be within partition bounds
[10:07] <ohsix> bummer
[10:08] <ohsix> there may be an alternate report, smartctl -x lists everything
[10:08] <smb> ppisati, In theory yes, though I am not 100% sure which boundaries are exactly checked at which level, butyes. Though the initial error was some dma write failing. And then some weird thing about invalid chs sector 0...
[10:09] <mIKEjONE2> hello, does anyone know how to use debian/rules to run make bzimage?
[10:10] <mIKEjONE2> after successfully building a kernel I wanted to make a few corrections
[10:10] <smb> cking, I am not really working, really. ;) I was just around to complain about jockey being broken now and the steam client steaming about it a bit. And apport not generating bug reports but only updates on errors.u.c which is hard to check for success or not
[10:10] <cking> ppisati, so you got hit by a link sequence error and I suspect then the SSD just popped offline
[10:11] <mIKEjONE2> but it seems as if build/rules binary-generic does not care about changes I make to .c files because make bzImage is never rerun
[10:13] <mIKEjONE2> it seems like runing build/rules clean ; build/rules binary-generic is a very poor way of recompiling the kernel, especially when I barely add a single printk
[10:15] <ppisati> cking: so more of a ATA bug then, but the ctrl communicates with the ssd so it could any of them actually
[10:15] <cking> ppisati, so it looks like an issue on the link, you got 10b8b and a LinkSeq error, so it looks like the SATA link got all weird
[10:16] <smb> mIKEjONE2, You may try to remove the build stamp file in debian/stamps (I believe) and skip the clean part
[10:16] <ppisati> cking: this is the haswell box, could be a bug in the ctrl then
[10:16] <ppisati> cking: or even silicon
[10:16] <smb> ppisati, Has that box a easy-swap drive bay like mine?
[10:17] <ppisati> smb: yes
[10:17] <smb> ppisati, You could replace the ssd by another disk to rule out the ssd
[10:17] <cking> ppisati, I got the same issue on Ivybridge with an Intel 520 SSD. So it may be SSD related or H/W related on the chipset, or both, or who knows
[10:17] <ppisati> i bet on "who knows"
[10:17] <cking> ppisati, but I didn't see the issue with the same SSD on a Core2 laptop
[10:17] <ppisati> smb: i'll try to update the fw, in cae it happens again i'll swap the disk
[10:18] <ppisati> (but it's nice to have a fast ssd... :) )
[10:18] <cking> ppisati, lemme rig up a spare Intel SSD on one of my SDP's and soak test it and see if we can characterise the bug
[10:18] <ppisati> when it doesn't crap out...
[10:18] <mIKEjONE2> so how do you guys recompile the kernel after modifying some of the source files?
[10:20] <mIKEjONE2> smb: that worked pretty well :) thanks
[10:20] <mIKEjONE2> I'm kind of surprised there's no better solution :/
[10:21] <smb> mIKEjONE2, We have a quick build box we share, so it is not really an issue, and recompiling the whole just makes sure everything really is done freshly and we are also a bit paranoid about that. ;)
[10:26] <mIKEjONE2> smb: I'm running this on a 6core i7 with 32GB of RAM + an SSD and it still takes 20mins :/
[10:31] <cking> mIKEjONE2, yep, 11 million lines of code, does take a while to build and package :-(
[10:31] <smb> there is always a good time to get another cup of coffee/tea or beer (depending on time of day) :)
[10:31] <cking> smb, I use build time to catch up on LKML
[10:32] <smb> cking, I knew I was doing something "wrong"...
[10:36] <mIKEjONE2> cking: well, not really, if you're building a bzimage without debian/rules framework, and you change a single file you don't have to rerun make mrproper and start from scratch
[10:37] <mIKEjONE2> make is smart enough to figure that out
[10:38] <mIKEjONE2> not quite sure why the debian/rules framework is intended for developers when its build capabilities are so crippled
[10:43] <cking> mIKEjONE2, use: rm debian/stamps/stamp-build-generic and then fakeroot debian/rules binary-generic, that will save a complete rebuild
[10:45] <cking> ppisati, which intel SSD do you have?
[10:45] <ppisati> cking: dmesg says
[10:45] <ppisati> [    4.066991] scsi 0:0:0:0: Direct-Access     ATA      INTEL SSDSC2CW24 400i PQ: 0 ANSI: 5
[10:46] <ppisati> 520?
[10:46] <ppisati> 400i is latest 520 fw
[10:46] <cking> same as mine 520
[10:46] <mIKEjONE2> cking: yea smb suggested that as well, and it works
[10:47] <mIKEjONE2> I'm just a little grumpy because I spent an hour on 3 rebuilds because I didn't know that trick :/
[10:47] <cking> :-(
[10:49] <mIKEjONE2> alternatively I just removed the touch $@ for the build-stamp rule from debian/rules.d/2-binary-arch.mk, this way I don't have to manually rm the file :D
[10:50] <cking> oh, that's way less of a hack, nice
[14:43] <Tassadar> Hi, I've sent patch to kernel-team@lists.ubuntu.com yesterday, but it did not show up in the archives. I am being told that I have to be subscribed to the mailing list, is that right?
[15:29] <diwic> is it expected that "linux-image-generic" is listed as one of the packages I can remove with 'apt-get autoremove'?
[16:03] <ppisati> and after an entire day of printk and debugging i finally found why we can poke at iomem space safely in Q (contrary to what happened to P)... ohhhhhhh....
[16:18] <cking> ppisati, nice
[16:38]  * ppisati calls it a day/week and heads to the gym for some ignorant weight lifting! :)