On Thu, 25 Feb 2016, Scott McGillivray wrote: > yes thank you. When using zram and -w 4096 it worked as expected. > > But when i re-tested with /dev/ram0 which is using the BLK_DEV_RAM/brd > ram device as cache. The make-bcache worked OK but when i tried to > register the device via "echo /dev/ram0 > /sys/fs/bcache/register" it > blows up with below error in dmesg. > > Call Trace: > [ 8460.086681] [<ffffffff812f8ae7>] kobj_attr_store+0xf/0x19 > [ 8460.086685] [<ffffffff811fd737>] sysfs_kf_write+0x3e/0x40 > [ 8460.086689] [<ffffffff811fcc40>] kernfs_fop_write+0xe9/0x136 > [ 8460.086694] [<ffffffff8119b0ed>] __vfs_write+0x28/0xa7 > [ 8460.086699] [<ffffffff81286ed3>] ? security_file_permission+0x3b/0x42 > [ 8460.086705] [<ffffffff8109b30c>] ? percpu_down_read+0x1f/0x41 > [ 8460.086708] [<ffffffff8119d604>] ? __sb_start_write+0x24/0x41 > [ 8460.086712] [<ffffffff8119bba2>] vfs_write+0x8f/0xe5 > [ 8460.086715] [<ffffffff8119c534>] SyS_write+0x46/0x79 > [ 8460.086720] [<ffffffff81645176>] entry_SYSCALL_64_fastpath+0x16/0x75 > [ 8460.086722] Code: 00 48 83 ff 04 77 09 49 89 bd b8 0a 00 00 eb 0f > e8 c6 c3 ff ff 48 ff c8 49 89 85 b8 0a 00 00 49 8b bd b8 0a 00 00 48 > 85 ff 75 02 <0f> 0b 48 ff c7 e8 a6 c3 ff ff 48 8d 14 c5 00 00 00 00 48 > ff c8 > [ 8460.086766] RIP [<ffffffffc00e832f>] register_bcache+0xe27/0x1423 [bcache] I just posted a patch in the thread "Re: bcache: register_cache() error opening sdh2" that might cleanup register_cache() error handling, and might give a better error---but ram0 devices are rather old and might not do what you want. I seem to remember they need to be initialized by dd if=/dev/zero of=/dev/ram0 or some such thing. I always seem to have trouble with /dev/ramX devices. You might be better off using /dev/loop0 and loopback a file in /dev/shm/ or /dev/hugepages/. -- Eric Wheeler, President eWheeler, Inc. dba Global Linux Security 888-LINUX26 (888-546-8926) Fax: 503-716-3878 PO Box 25107 www.GlobalLinuxSecurity.pro Linux since 1996! Portland, OR 97298 > > > > On 25 February 2016 at 15:39, Eric Wheeler <bcache@xxxxxxxxxxxxxxxxxx> wrote: > > On Thu, 25 Feb 2016, Eric Wheeler wrote: > > > >> On Thu, 25 Feb 2016, Scott McGillivray wrote: > >> > >> > that doesn't work either. It gives the same error from my opening > >> > post. Although this time it's also giving a new message in dmesg about > >> > block size. > >> > > >> > > >> > bcache: bch_cached_dev_attach() Couldn't attach sdf: block size less > >> > than set's block size > >> > bcache: __cached_dev_store() Can't attach 0bba8852-ad4f-41f4-bfa3-679a886a14ad > >> > : cache set not found > >> > > >> > which is strange because both devices block size are the same > >> > > >> > $ blockdev --getbsz /dev/sdf > >> > 4096 > >> > > >> > $ blockdev --getbsz /dev/zram0 > >> > 4096 > >> > >> > >> Try this: > >> make-bcache -w 4096 -C /dev/zram0 > > > > I'm not sure if -w needs to be called on the backing device too---but if > > you do then expect to loose data when reformatting. > > > > If you decide reformat both backing and cache devices to test, then you > > might consider aligning your data to your backing volume's stride width > > with --data-offset NNN while you're at it if you haven't already. > > > > -Eric > > > >> > >> > > >> > > >> > On 25 February 2016 at 14:44, Eric Wheeler <bcache@xxxxxxxxxxxxxxxxxx> wrote: > >> > > On Thu, 25 Feb 2016, Scott McGillivray wrote: > >> > > > >> > >> No, doesn't' work either i'm afraid. Below is example of failing > >> > >> silently with ram disk and succeeded with HDD as cache device. > >> > >> > >> > >> Tested with /lib/modules/4.5.0-rc3+/kernel/drivers/md/bcache/bcache.ko > >> > >> > >> > >> > >> > >> with zram - doesn't work > >> > >> ================= > >> > >> > >> > >> $ ls -al /sys/fs/bcache/ > >> > >> total 0 > >> > >> drwxr-xr-x 2 root root 0 Feb 25 13:42 . > >> > >> drwxr-xr-x 9 root root 0 Feb 25 13:40 .. > >> > >> --w------- 1 root root 4096 Feb 25 13:40 register > >> > >> --w------- 1 root root 4096 Feb 25 13:40 register_quiet > >> > >> > >> > >> $ file /dev/zram0 > >> > >> /dev/zram0: block special (250/0) > >> > >> > >> > >> $ fdisk -l /dev/zram0 > >> > >> Disk /dev/zram0: 1 GiB, 1073741824 bytes, 262144 sectors > >> > >> > >> > >> $ make-bcache -C /dev/zram0 > >> > >> UUID: 87e63981-7bd8-4cc3-93db-9ca275367c21 > >> > >> Set UUID: 0bba8852-ad4f-41f4-bfa3-679a886a14ad > >> > >> version: 0 > >> > >> nbuckets: 2048 > >> > >> block_size: 8 > >> > >> bucket_size: 1024 > >> > >> nr_in_set: 1 > >> > >> nr_this_dev: 0 > >> > >> first_bucket: 1 > >> > > > >> > > did you `echo /dev/zram0 > /sys/fs/bcache/register` > >> > > and then `echo <CSET-UUID> > /sys/block/bcache0/bcache/attach` > >> > > > >> > > Not sure which UUID is the CSET UUID. Probably > >> > > 0bba8852-ad4f-41f4-bfa3-679a886a14ad. > >> > > > >> > > See attaching bcache0 to a cache: > >> > > https://www.kernel.org/doc/Documentation/bcache.txt > >> > > > >> > > -Eric > >> > > > >> > >> > >> > >> > >> > >> No errors, but nothing in dmesg and contents of /sys/fs/bcache/ remain > >> > >> exactly the same... no UUID. Seems to just silently fail. > >> > >> > >> > >> > >> > >> with HDD - works as expected. > >> > >> ===================== > >> > >> > >> > >> $ fdisk -l /dev/sdf > >> > >> Disk /dev/sdf: 1 GiB, 1073741824 bytes, 2097152 sectors > >> > >> > >> > >> $ file /dev/sdf > >> > >> /dev/sdf: block special (8/80) > >> > >> > >> > >> $ make-bcache -C /dev/sdf > >> > >> UUID: 3234c9e7-8186-4a7c-99a8-ae6a371c6f2c > >> > >> Set UUID: 9a71faf7-a408-4f4d-906e-e5d67e43291f > >> > >> version: 0 > >> > >> nbuckets: 2048 > >> > >> block_size: 1 > >> > >> bucket_size: 1024 > >> > >> nr_in_set: 1 > >> > >> nr_this_dev: 0 > >> > >> first_bucket: 1 > >> > >> > >> > >> $ ls -al /sys/fs/bcache/ > >> > >> total 0 > >> > >> drwxr-xr-x 3 root root 0 Feb 25 13:48 . > >> > >> drwxr-xr-x 9 root root 0 Feb 25 13:40 .. > >> > >> drwxr-xr-x 7 root root 0 Feb 25 13:48 9a71faf7-a408-4f4d-906e-e5d67e43291f > >> > >> --w------- 1 root root 4096 Feb 25 13:40 register > >> > >> --w------- 1 root root 4096 Feb 25 13:40 register_quiet > >> > >> > >> > >> $ dmesg > >> > >> [ 714.168488] bcache: register_cache() registered cache device sdf > >> > >> > >> > >> > >> > >> Thank you. > >> > >> > >> > >> On 24 February 2016 at 15:37, Eric Wheeler <lists@xxxxxxxxxxxx> wrote: > >> > >> > > >> > >> > On Tue, 16 Feb 2016, Scott McGillivray wrote: > >> > >> > > >> > >> >> Hello > >> > >> >> > >> > >> >> I am using Ubuntu 15.10 with latest nightly 4.5 kernel. Ubuntu creates > >> > >> >> 16 RAM drives as /dev/ram0 - ram16 by default and i am trying to test > >> > >> >> bcache and use one of these as a cache device. > >> > >> >> > >> > >> >> wipefs /dev/sdd > >> > >> >> wipefs /dev/sdd -a > >> > >> >> wipefs /dev/ram0 > >> > >> >> wipefs /dev/ram0 -a > >> > >> >> make-bcache -B /dev/sdd -C /dev/ram0 > >> > >> > > >> > >> > Does it work with /dev/zram0? You would prep it something like this: > >> > >> > > >> > >> > modprobe zram > >> > >> > echo $((1024*1024*1024)) > /sys/block/zram0/disksize # 1GB > >> > >> > make-bcache -C /dev/zram0 > >> > >> > > >> > >> > -Eric > >> > >> > > >> > >> > > >> > >> > > >> > >> >> At this point there were no errors output but there is also no UUID > >> > >> >> entry in /sys/fs/bcache/. Only register and register_quiet. > >> > >> >> > >> > >> >> bache-super-show correctly prints info about both block and cache > >> > >> >> devices when queried but when i try to manually attach the cache > >> > >> >> device with echo cset.uuid > /sys/block/bcache0/bcache/attach i get no > >> > >> >> error but in dmesg i see > >> > >> >> > >> > >> >> bcache: __cached_dev_store() Can't attach > >> > >> >> 3f2521a6-2103-4e2e-a47d-cc56558e8f7d: cache set not found > >> > >> >> > >> > >> >> If i repeat this example but use another sd block device as cache then > >> > >> >> it works OK. > >> > >> >> > >> > >> >> Can anyone confirm if it should be possible to use the ramdisk based > >> > >> >> block device as a caching device with bcache ? > >> > >> >> > >> > >> >> I also tried to use a loop device as cache and that didn't work > >> > >> >> either.. i got "kernel bug" in dmesg for > >> > >> >> drivers/md/bcache/super.c:1812. RIP [] > >> > >> >> cache_alloc.isra.21+0x646/0x670[bcache] > >> > >> >> > >> > >> >> Does bcache need the cache device to be a certain block type i.e sd 8 > >> > >> >> ? Should i be able to use any block device from /proc/devices as a > >> > >> >> cache device ? > >> > >> >> > >> > >> >> The kernel options used for ramdisk are: > >> > >> >> > >> > >> >> CONFIG_BLK_DEV_RAM=y > >> > >> >> CONFIG_BLK_DEV_RAM_COUNT=16 > >> > >> >> CONFIG_BLK_DEV_RAM_SIZE=65536 > >> > >> >> CONFIG_BLK_DEV_RAM_DAX=y > >> > >> >> > >> > >> >> Thank you. > >> > >> >> Scott > >> > >> >> -- > >> > >> >> 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 > >> > >> >> > >> > >> > > >> > >> > > >> > >> > > >> > >> > -- > >> > >> > Eric Wheeler, President eWheeler, Inc. dba Global Linux Security > >> > >> > 888-LINUX26 (888-546-8926) Fax: 503-716-3878 PO Box 25107 > >> > >> > www.GlobalLinuxSecurity.pro Linux since 1996! Portland, OR 97298 > >> > >> > > >> > >> -- > >> > >> 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 > >> > >> > >> > > >> -- > >> 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 > >> > -- > 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 > -- 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
