Date: Tue, 5 Dec 2000 18:29:12 -0500 From: Barry Lustig <barry@lustig.com> To: stable@freebsd.org, Matt Dillon <dillon@earth.backplane.com> Subject: VMware hanging Message-ID: <20001205232913.15317.qmail@devious.lustig.com> In-Reply-To: <200011192324.eAJNO1891133@earth.backplane.com> References: <20001118231633.A85206@evilcode.com> <200011190635.eAJ6ZRS94516@grumpy.dyndns.org> <20001118230305.A83848@evilcode.com> <20001119151138.A7434@bsdvm.jtjang.idv.tw> <200011192202.eAJM2cG03593@billy-club.village.org> <200011192239.eAJMdDa90818@earth.backplane.com> <200011192308.eAJN8q714360@whizzo.transsys.com> <200011192324.eAJNO1891133@earth.backplane.com>
next in thread | previous in thread | raw e-mail | index | archive | help
VMware is hanging the system in certain cases with 4.2-STABLE kernels, but
works with 4.2-RELEASE. If I config the memory in vmware to 64MB the client
os runs properly. If I set it higher, I get a hung system. When the system
is hung everything seems to lock up except for the mouse.
At the recommendation of Doug Ambrisko I started building kernels with
4.2-RELEASE sources and added commits one at a time. The place where the
hang starts is after adding in: dillon 2000/11/25 18:55:15 PST (See
below)
Just as a sanity check I did:
cvs co -rRELENG_4 -D"2000/11/25 18:35:00 PST" sys
and got a working kernel. When I rm -rf /usr/src/sys and did:
cvs co -rRELENG_4 -D"2000/11/25 19:00:00 PST" sys
I get the kernel that hangs. I have my system running off of a serial
console so I can get information that way. I'd like to help track this down
but am not sure how to proceed. Any suggestions?
barry
dillon 2000/11/25 18:55:15 PST
Modified files: (Branch: RELENG_4)
sys/kern vfs_bio.c vfs_cluster.c vfs_subr.c
vfs_vnops.c
sys/sys buf.h vnode.h
sys/ufs/ffs ffs_inode.c ffs_softdep.c
sys/ufs/ufs ufs_readwrite.c
sys/vm swap_pager.c vm_page.c vm_page.h
vm_pageout.c
Log:
MFC the major low memory deadlock solution. This patch removes most of
the preexisting low-memory-handling hacks and replaces it with a relatively
simple concept: The entire I/O path (except networking) is allowed to dig
into the system memory reserve but will effectively free resources rather
then leave them wired/cached if the system is found to be in a low-memory
situation, thus allowing I/O to continue to operate.
This patch also fixes a VXLOCK self-deadlock and generates a message
if it detects what used to be the deadlock. (this also MFCd)
Finally, this patch removes code related to the case where the original
page could not be found during bogus page recovery and replaces it with a
panic. The original page had better still exist since we have an extra
reference on it and on the VM object during the I/O. (this also MFCd)
Revision Changes Path
1.242.2.4 +92 -62 src/sys/kern/vfs_bio.c
1.92.2.2 +7 -1 src/sys/kern/vfs_cluster.c
1.249.2.7 +11 -5 src/sys/kern/vfs_subr.c
1.87.2.4 +4 -2 src/sys/kern/vfs_vnops.c
1.88.2.2 +2 -1 src/sys/sys/buf.h
1.111.2.3 +2 -1 src/sys/sys/vnode.h
1.56.2.1 +4 -1 src/sys/ufs/ffs/ffs_inode.c
1.57.2.5 +20 -18 src/sys/ufs/ffs/ffs_softdep.c
1.65.2.3 +5 -1 src/sys/ufs/ufs/ufs_readwrite.c
1.130.2.7 +4 -2 src/sys/vm/swap_pager.c
1.147.2.4 +30 -7 src/sys/vm/vm_page.c
1.75.2.4 +3 -1 src/sys/vm/vm_page.h
1.151.2.5 +98 -74 src/sys/vm/vm_pageout.c
To Unsubscribe: send mail to majordomo@FreeBSD.org
with "unsubscribe freebsd-stable" in the body of the message
Want to link to this message? Use this URL: <https://mail-archive.FreeBSD.org/cgi/mid.cgi?20001205232913.15317.qmail>