On Sun, Apr 05, 2015 at 08:37:23PM +0300, Konstantin Belousov wrote: > On Sun, Apr 05, 2015 at 06:47:21PM +0300, Gleb Smirnoff wrote: > > This is r281079. > > > > Since vm_page_advise() may call vm_page_dirty() in the MADV_DONTNEED case, > > the assertion is valid. So, looks like vm_fault_dontneed() needs W-lock on > > the first_object. > > > Either this, or vm_page_advise() could be changed to do nothing in the > case of MADV_DONTNEED operation and a need to call vm_page_dirty(), if > the object is only read-locked.
I have since managed to build & (successfully) boot: s3:FreeBSD g1-254.catwhisker.org 11.0-CURRENT FreeBSD 11.0-CURRENT #0 r281106M/281106:1100067: Sun Apr 5 09:57:39 PDT 2015 r...@g1-254.catwhisker.org:/common/S3/obj/usr/src/sys/CANARY amd64 So I'm about to try re-building the i386 head (and be sure that it's quite "clean" in the process). (I had seen the panic for i386 twice out of two attempts, so I'm reasonably certain it wasn't a fluke. But it could be something "weird" in my head/i386 environment, as that is the only one where I've had a problem with today's builds & smoke-tests. I'm using both i386 & amd64, and both stable/10 and head -- 4 combinations; one for each of 4 slices on the same hardware.) Peace, david -- David H. Wolfskill da...@catwhisker.org Those who murder in the name of God or prophet are blasphemous cowards. See http://www.catwhisker.org/~david/publickey.gpg for my public key.
Description: PGP signature