Hi Johannes, On Mon, Aug 06, 2018 at 03:27:25PM +0200, Johannes Thumshirn wrote: > On Mon, Aug 06, 2018 at 03:22:34PM +0200, ard wrote: > > Ahh, yeah, sorry. > > I did test them. > > I can't say I see a regression. > > I also can't say I see progress, although it feels like there is > > a tad less leaking. > > For that I need to do some statistics. > > > > But for now I ack that it does not regress further. > > From what I've seen in your kmemleak reports, it wasn't only > libfc/fcoe that was leaking memory. > > So kmemleak reports are welcome (especially if they confirm libfc/fcoe > is clean now ;-)). I've updated the ticket: https://github.com/hardkernel/linux/issues/360#issuecomment-410721997 dmesg (with full debugging): https://github.com/hardkernel/linux/files/2262858/dmesg-2018-08-06.txt and kmemleak: https://github.com/hardkernel/linux/files/2262861/kmemleak-2018-08-06.txt I'll have to take a good look at the what and why. I'm gonna git clean -x -d and rebuild the kernel, because this is weird. This is the tree I build: https://github.com/ardje/linux/tree/gb-4.14-fcoe-patch So vanilla linux-4.14 with the patch set: https://github.com/ardje/linux/commit/747bf04057a99be5b01f768654cfd61bc9f4fc6c Meh, I should try to use git for patching. I will spam some printk's later on, just to check which flows are followed (if the git clean doesn't reveal some problem, although I can't imagine). Regards, Ard -- .signature not found