On 23 ноября 2012 09:51:39 Stefan Kalkowski wrote:

Hi,

thanks for detailed bug report. The problem isn't located in L4Linux nor in its block driver, but in the ATAPI driver. The creation of an EXT filesystem resulted in: the Linux kernel requesting a lot of blocks at once from the ATAPI driver. When the driver initiated DMA operations, it doesn't respect the limit of blocks that is possible to transfer via DMA. Thereby, the driver stucked when getting no further interrupts, and the Linux block-layer stucked too.

I've described this behaviour in an issue in our issue-tracker:

https://github.com/genodelabs/genode/issues/514

To solve the problem, you might try the following patch:

https://github.com/genodelabs/genode/commit/1ab32273d25646c33a69b469e8756ed7 84918bf0.patch

or use the staging branch of the Genode repository.

Regards Stefan

On 11/20/2012 11:33 AM, Timur wrote:

...

Hi! mke2fs with simple block device (l4linux + atapi drv) work's correctly, but if i create encrypted device with cryptsetup, open it (there device mapper file created) and try to create file system (on dm-device) using mke2fs then system freezes(infinitely write to device). I found that problem in fsync call at unix_flush(io_channel channel) function (mke2fs_source/lib/ext2fs/unix_io.c). Help me to solve this problem) .run file attached initrd with cryptsetup - https://github.com/shomich/files/blob/master/initrd.gz

---

---- Monitor your physical, virtual and cloud infrastructure from a single web console. Get in-depth insight into apps, servers, databases, vmware, SAP, cloud infrastructure, etc. Download 30-day Free Trial. Pricing starts from $795 for 25 servers or applications! http://p.sf.net/sfu/zoho_dev2dev_nov

---

Genode-main mailing list Genode-main@lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/genode-main

Thanks, but I've already test it and it's doesn't work with device-mapper device

Hello,

I've to admit, that the current staging branch seems to be a bit unstable with regard to L4Linux. I realized a hang of the Linux kernel while booting. But it's hard to reproduce, and didn't occured after the login is available.

in the meanwhile, the fix went to genodelabs/master.

Generally, I would not recommend to use the staging branch because the commits that are not already merged into master may be rebased at any time. When the automated build and test bot detects a flaw in one of the pending patches in staging, those patches will be updated.

Instead of using staging, I recommend to checkout the master branch into a private branch and apply individual commits from other branches. This is quite convenient using 'git cherry-pick'.

Cheers Norman