On Mon, 2014-01-06 at 15:37 -0800, Kent Overstreet wrote: +AD4- On Fri, Dec 20, 2013 at 03:46:30PM +-0000, Chris Mason wrote: +AD4- +AD4- On Fri, 2013-12-20 at 10:42 -0200, F+AOE-bio Pfeifer wrote: +AD4- +AD4- +AD4- Hello, +AD4- +AD4- +AD4- +AD4- +AD4- +AD4- I put the +ACI-WARN+AF8-ON(1)+ADsAIg- after the printk lines (incomplete page read +AD4- +AD4- +AD4- and incomplete page write) in extent+AF8-io.c. +AD4- +AD4- +AD4- +AD4- +AD4- +AD4- here some call traces: +AD4- +AD4- +AD4- +AD4- +AD4- +AD4- +AFs- 19.509497+AF0- incomplete page read in btrfs with offset 2560 and length 1536 +AD4- +AD4- +AD4- +AFs- 19.509500+AF0- ------------+AFs- cut here +AF0------------- +AD4- +AD4- +AD4- +AFs- 19.509528+AF0- WARNING: CPU: 2 PID: 220 at fs/btrfs/extent+AF8-io.c:2441 +AD4- +AD4- +AD4- end+AF8-bio+AF8-extent+AF8-readpage+-0x788/0xc20 +AFs-btrfs+AF0-() +AD4- +AD4- +AD4- +AFs- 19.509530+AF0- Modules linked in: cdc+AF8-acm fuse iTCO+AF8-wdt +AD4- +AD4- +AD4- iTCO+AF8-vendor+AF8-support snd+AF8-hda+AF8-codec+AF8-analog coretemp kvm+AF8-intel kvm raid1 +AD4- +AD4- +AD4- ext4 crc16 md+AF8-mod mbcache jbd2 microcode nvidia(PO) psmouse pcspkr +AD4- +AD4- +AD4- evdev serio+AF8-raw i2c+AF8-i801 lpc+AF8-ich i2c+AF8-core snd+AF8-hda+AF8-intel sky2 skge +AD4- +AD4- +AD4- i82975x+AF8-edac button asus+AF8-atk0110 snd+AF8-hda+AF8-codec snd+AF8-hwdep shpchp +AD4- +AD4- +AD4- snd+AF8-pcm snd+AF8-page+AF8-alloc snd+AF8-timer acpi+AF8-cpufreq snd edac+AF8-core soundcore +AD4- +AD4- +AD4- processor vboxdrv(O) sr+AF8-mod cdrom ata+AF8-generic pata+AF8-acpi hid+AF8-generic +AD4- +AD4- +AD4- usbhid hid usb+AF8-storage sd+AF8-mod pata+AF8-marvell firewire+AF8-ohci uhci+AF8-hcd ahci +AD4- +AD4- +AD4- ehci+AF8-pci firewire+AF8-core ata+AF8-piix libahci crc+AF8-itu+AF8-t ehci+AF8-hcd libata +AD4- +AD4- +AD4- scsi+AF8-mod usbcore usb+AF8-common btrfs crc32c libcrc32c xor raid6+AF8-pq bcache +AD4- +AD4- +AD4- +AFs- 19.509578+AF0- CPU: 2 PID: 220 Comm: btrfs-endio-met Tainted: P +AD4- +AD4- +AD4- W O 3.12.5-1-ARCH +ACM-1 +AD4- +AD4- +AD4- +AFs- 19.509580+AF0- Hardware name: System manufacturer System Product +AD4- +AD4- +AD4- Name/P5WDG2 WS Pro, BIOS 0905 03/06/2008 +AD4- +AD4- +AD4- +AFs- 19.509581+AF0- 0000000000000009 ffff880231a63cb0 ffffffff814ee37b +AD4- +AD4- +AD4- 0000000000000000 +AD4- +AD4- +AD4- +AFs- 19.509585+AF0- ffff880231a63ce8 ffffffff81062bcd ffffea00085eaec0 +AD4- +AD4- +AD4- 0000000000000000 +AD4- +AD4- +AD4- +AFs- 19.509587+AF0- ffff8802320cc9c0 0000000000000000 ffff880233b0e000 +AD4- +AD4- +AD4- ffff880231a63cf8 +AD4- +AD4- +AD4- +AFs- 19.509590+AF0- Call Trace: +AD4- +AD4- +AD4- +AFs- 19.509596+AF0- +AFsAPA-ffffffff814ee37b+AD4AXQ- dump+AF8-stack+-0x54/0x8d +AD4- +AD4- +AD4- +AFs- 19.509601+AF0- +AFsAPA-ffffffff81062bcd+AD4AXQ- warn+AF8-slowpath+AF8-common+-0x7d/0xa0 +AD4- +AD4- +AD4- +AFs- 19.509603+AF0- +AFsAPA-ffffffff81062caa+AD4AXQ- warn+AF8-slowpath+AF8-null+-0x1a/0x20 +AD4- +AD4- +AD4- +AFs- 19.509614+AF0- +AFsAPA-ffffffffa00b7ba8+AD4AXQ- end+AF8-bio+AF8-extent+AF8-readpage+-0x788/0xc20 +AFs-btrfs+AF0- +AD4- +AD4- +AD4- +AD4- This should mean that bcache is either failing to read some blocks +AD4- +AD4- properly or is fiddling with the bv+AF8-len/bv+AF8-offset fields. +AD4- +AD4- +AD4- +AD4- Could someone from bcache comment? +AD4- +AD4- Oh man, I found this and then threw up my hands in despair. +AD4- +AD4- Bcache isn't doing anything with the bv+AF8-len/bv+AF8-offset fields+ADs- it may clone the +AD4- biovec so it can retry a bio on error, if the biovecs weren't all whole pages, +AD4- otherwise it just passes the biovec down with the next bio to the underlying +AD4- cache/backing device. +AD4- +AD4- What btrfs appears to be doing though - I couldn't believe that code actually +AD4- +AF8-worked+AF8-, Jens please jump in here but AFAIK bv+AF8-len/bv+AF8-offset are in practice +AD4- undefined after a bio's completed, they might have been updated if the driver +AD4- was using blk+AF8-update+AF8-request but for many drivers that just process the entire +AD4- bio all at once they just won't touch those fields - and that includes anything +AD4- that clones the bio (md/dm). +AD4- +AD4- This is probably relevant to immutable biovecs here... +AD4- +AD4- ------------- +AD4- +AD4- Ok, I looked again at the relevant btrfs code, I guess I can see how this printk +AD4- isn't normally triggered. But Chris, +AF8-what on earth+AF8- is btrfs trying to check +AD4- for here? And why is it using bv+AF8-offset and bv+AF8-len further down in +AD4- end+AF8-bio+AF8-extent+AF8-readpage()? After the IO is done, we're recording the specific logical byte range that covered the IO. In practice its always the full page, we can switch to just trusting PAGE+AF8-CACHE+AF8-SIZE. -chris -- To unsubscribe from this list: send the line "unsubscribe linux-bcache" in the body of a message to majordomo@xxxxxxxxxxxxxxx More majordomo info at http://vger.kernel.org/majordomo-info.html
