kvm-56 release

This restores live migration support which has been broken for a couple of releases. Also, if you had Java problems please test kvm-56, as it has some nx related permission fixes.

Changes from kvm-55:
- Generalize exception handling mechanism
- fix x86 emulator eflags handling
- avoid exit to userspace on cr8 write if using kernel lapic on Intel
- fix stack instructions emulation in 64-bit mode
- fixes Vista x86
- except on Woodcrests
- trap cr8 access on amd
- mmu simplification
- fix setting nx bit on shadow pte
- fix inherited mmu permissions
- more portability work (Hollis Blanchard, Xiantao Zhang)
- add cr8 to 'mov cr' instruction emulation (Joerg Roedel)
- exit to userspace on cr8 write if not kernel apic on amd (Joerg Roedel)
- atomically set accessed and dirty bit on guest ptes (Marcelo Tosatti)
- use ->fault() vma op instead of ->nopage()
- remove abi 10 libkvm compatibility
- add some x86 emulator tests
- add many mmu access tests
- fix rpm build on Fedora 8 (Dan Kenigsberg)
- fix 'make sync' on debian and derivatives (Ryan Harper)
- fix live migration (Uri Lublin, Izik Eidus)
- initial ia64 userspace (Xiantao Zhang)

Notes:
If you use the modules bundled with kvm-56, you can use any version
of Linux from 2.6.9 upwards.
If you use the modules bundled with Linux 2.6.20, you need to use
kvm-12.
If you use the modules bundled with Linux 2.6.21, you need to use
kvm-17.
Modules from Linux 2.6.22 and up will work with any kvm version from
kvm-22. Some features may only be available in newer releases.
For best performance, use Linux 2.6.23-rc2 or later as the host.

http://kvm.qumranet.com

-------------------------------------------------------------------------
SF.Net email is sponsored by:
Check out the new SourceForge.net Marketplace.
It's the best place to buy or sell services
for just about anything Open Source.
http://ad.doubleclick.net/clk;164216239;13503038;w?http://sf.net/marketplace

I get the following error when modprobing the new module:

WARNING: Error inserting kvm (/lib/modules/2.6.22-14-generic/extra/kvm.ko): Unknown symbol in module, or unknown parameter (see dmesg)
kvm: Unknown symbol cmpxchg64

Ubuntu 7.10 32-bits, C2D E6300

Pelle.

-------------------------------------------------------------------------
SF.Net email is sponsored by:
Check out the new SourceForge.net Marketplace.
It's the best place to buy or sell services
for just about anything Open Source.
http://ad.doubleclick.net/clk;164216239;13503038;w?http://sf.net/marketplace

-----Original Message-----
From: kvm-devel-bounces-5NWGOfrQmneRv+LV9MX5uipxlwaOVQ5f@public.gmane.org on behalf of Pelle
Sent: Thu 12/13/2007 6:19 AM
To: Avi Kivity
Cc: kvm-devel
Subject: Re: [kvm-devel] [ANNOUNCE] kvm-56 release

I hope you did rmmod first. You need to do:

rmmod kvm-intel
rmmod kvm
modprobe kvm-intel

-Alexey "Technologov"

-------------------------------------------------------------------------
SF.Net email is sponsored by:
Check out the new SourceForge.net Marketplace.
It's the best place to buy or sell services
for just about anything Open Source.
http://ad.doubleclick.net/clk;164216239;13503038;w?http://sf.net/marketplace

_______________________________________________
kvm-devel mailing list [email protected]
https://lists.sourceforge.net/lists/listinfo/kvm-devel

Yes, unfortunately that's not it.

-------------------------------------------------------------------------
SF.Net email is sponsored by:
Check out the new SourceForge.net Marketplace.
It's the best place to buy or sell services
for just about anything Open Source.
http://ad.doubleclick.net/clk;164216239;13503038;w?http://sf.net/marketplace

Please try the attached patch.

error compiling committee.c: too many arguments to function

index ea94f30..b044cf9 100644
@@ -322,6 +322,28 @@ static inline void __kvm_set_64bit(u64 *ptr, u64 val)
 #undef  set_64bit
 #define set_64bit __kvm_set_64bit
 
+static inline unsigned long long __kvm_cmpxchg64(volatile void *ptr,
+						 unsigned long long old,
+						 unsigned long long new)
+{
+	unsigned long long prev;
+	__asm__ __volatile__(LOCK_PREFIX "cmpxchg8b %3"
+			     : "=A"(prev)
+			     : "b"((unsigned long)new),
+			       "c"((unsigned long)(new >> 32)),
+			       "m"(*__xg(ptr)),
+			       "0"(old)
+			     : "memory");
+	return prev;
+}
+
+#define kvm_cmpxchg64(ptr,o,n)\
+	((__typeof__(*(ptr)))__kvm_cmpxchg64((ptr),(unsigned long long)(o),\
+					(unsigned long long)(n)))
+
+#undef cmpxchg64
+#define cpmxchg64(ptr, o, n) kvm_cmpxchg64(ptr, o, n)
+
 #endif
 
 #ifndef CONFIG_PREEMPT_NOTIFIERS

-------------------------------------------------------------------------
SF.Net email is sponsored by:
Check out the new SourceForge.net Marketplace.
It's the best place to buy or sell services
for just about anything Open Source.
http://ad.doubleclick.net/clk;164216239;13503038;w?http://sf.net/marketplace

_______________________________________________
kvm-devel mailing list [email protected]
https://lists.sourceforge.net/lists/listinfo/kvm-devel

Patch applied, rebuild (twice) but still the same error.

-------------------------------------------------------------------------
SF.Net email is sponsored by:
Check out the new SourceForge.net Marketplace.
It's the best place to buy or sell services
for just about anything Open Source.
http://ad.doubleclick.net/clk;164216239;13503038;w?http://sf.net/marketplace

Sorry, there was a type in the patch. Please change cpmxchg64 to cmpxchg64 in kernel/external-module-compat.h, or apply the attached corrected patch on a clean kvm-56.

error compiling committee.c: too many arguments to function

index ea94f30..1dc5a96 100644
@@ -322,6 +322,28 @@ static inline void __kvm_set_64bit(u64 *ptr, u64 val)
 #undef  set_64bit
 #define set_64bit __kvm_set_64bit
 
+static inline unsigned long long __kvm_cmpxchg64(volatile void *ptr,
+						 unsigned long long old,
+						 unsigned long long new)
+{
+	unsigned long long prev;
+	__asm__ __volatile__(LOCK_PREFIX "cmpxchg8b %3"
+			     : "=A"(prev)
+			     : "b"((unsigned long)new),
+			       "c"((unsigned long)(new >> 32)),
+			       "m"(*__xg(ptr)),
+			       "0"(old)
+			     : "memory");
+	return prev;
+}
+
+#define kvm_cmpxchg64(ptr,o,n)\
+	((__typeof__(*(ptr)))__kvm_cmpxchg64((ptr),(unsigned long long)(o),\
+					(unsigned long long)(n)))
+
+#undef cmpxchg64
+#define cmpxchg64(ptr, o, n) kvm_cmpxchg64(ptr, o, n)
+
 #endif
 
 #ifndef CONFIG_PREEMPT_NOTIFIERS

-------------------------------------------------------------------------
SF.Net email is sponsored by:
Check out the new SourceForge.net Marketplace.
It's the best place to buy or sell services
for just about anything Open Source.
http://ad.doubleclick.net/clk;164216239;13503038;w?http://sf.net/marketplace

_______________________________________________
kvm-devel mailing list [email protected]
https://lists.sourceforge.net/lists/listinfo/kvm-devel

Ah, check.

Thank you.

-------------------------------------------------------------------------
SF.Net email is sponsored by:
Check out the new SourceForge.net Marketplace.
It's the best place to buy or sell services
for just about anything Open Source.
http://ad.doubleclick.net/clk;164216239;13503038;w?http://sf.net/marketplace

Hi Pelle!

Yes, You're right, KVM-56 doesn't compiles on 32-bit systems. (tested on openSUSE 10.3, 32-bit)

-Alexey Technologov

-------------------------------------------------------------------------
SF.Net email is sponsored by:
Check out the new SourceForge.net Marketplace.
It's the best place to buy or sell services
for just about anything Open Source.
http://ad.doubleclick.net/clk;164216239;13503038;w?http://sf.net/marketplace

_______________________________________________
kvm-devel mailing list [email protected]
https://lists.sourceforge.net/lists/listinfo/kvm-devel

It compiles on 32-bit systems. It doesn't compile on 32-bit non-pae
systems.

error compiling committee.c: too many arguments to function


-------------------------------------------------------------------------
SF.Net email is sponsored by:
Check out the new SourceForge.net Marketplace.
It's the best place to buy or sell services
for just about anything Open Source.
http://ad.doubleclick.net/clk;164216239;13503038;w?http://sf.net/marketplace

gentoo non-pae kernels build ok as explained in previous post when using :

CONFIG_NOHIGHMEM=y
CONFIG_X86_PAE not set

enabling HIGHMEM64G and PAE also enables CONFIG_X86_CMPXCHG64=y

Carlo

-------------------------------------------------------------------------
SF.Net email is sponsored by:
Check out the new SourceForge.net Marketplace.
It's the best place to buy or sell services
for just about anything Open Source.
http://ad.doubleclick.net/clk;164216239;13503038;w?http://sf.net/marketplace

it compiles ok in Gentoo 2007.0 building against the last stable gentoo kernels 2.6.22 and 2.6.23 for x86 :

gentoo-sources-2.6.22-r9
gentoo-sources-2.6.23-r3

Carlo

PS. 2.6.22 is missing a declaration for cmpxchg64 as reported by Pele.

-------------------------------------------------------------------------
SF.Net email is sponsored by:
Check out the new SourceForge.net Marketplace.
It's the best place to buy or sell services
for just about anything Open Source.
http://ad.doubleclick.net/clk;164216239;13503038;w?http://sf.net/marketplace

unfortunately 32bit centos guest still not be able to boot on 64bit host.

  Levente                               "Si vis pacem para bellum!"

-------------------------------------------------------------------------
SF.Net email is sponsored by:
Check out the new SourceForge.net Marketplace.
It's the best place to buy or sell services
for just about anything Open Source.
http://ad.doubleclick.net/clk;164216239;13503038;w?http://sf.net/marketplace

Farkas, you cannot boot the guest with single cpu?
i am working now on a fix for the smp problem with RHEL 5 32bits, but it always
work for me when i try to boot it with single cpu (i tried it on servel computers)

-------------------------------------------------------------------------
SF.Net email is sponsored by:
Check out the new SourceForge.net Marketplace. It's the best place to buy or sell services for just about anything Open Source.
http://ad.doubleclick.net/clk;164216239;13503038;w?http://sf.net/marketplace

sorry i forget to mention the boot only hangs with smp.

  Levente                               "Si vis pacem para bellum!"

-------------------------------------------------------------------------
SF.Net email is sponsored by:
Check out the new SourceForge.net Marketplace.
It's the best place to buy or sell services
for just about anything Open Source.
http://ad.doubleclick.net/clk;164216239;13503038;w?http://sf.net/marketplace

can you please try to convert the image to vmdk and run it with -smp 2 -no-kvm-irqchip and tell me what happen?

-------------------------------------------------------------------------
SF.Net email is sponsored by:
Check out the new SourceForge.net Marketplace.
It's the best place to buy or sell services
for just about anything Open Source.
http://ad.doubleclick.net/clk;164216239;13503038;w?http://sf.net/marketplace

BTW, there were changes between 2.6.24-rc3 and -rc5 such that there are merge conflicts when I tried to do a git pull from kvm/master unto Linus's git mainline.

- Ted

-------------------------------------------------------------------------
SF.Net email is sponsored by:
Check out the new SourceForge.net Marketplace.
It's the best place to buy or sell services
for just about anything Open Source.
http://ad.doubleclick.net/clk;164216239;13503038;w?http://sf.net/marketplace

In general I merge regularly and resolve conflicts. Currently kvm.git is based on 2.6.24-rc5.

Any sufficiently difficult bug is indistinguishable from a feature.


-------------------------------------------------------------------------
SF.Net email is sponsored by:
Check out the new SourceForge.net Marketplace.
It's the best place to buy or sell services
for just about anything Open Source.
http://ad.doubleclick.net/clk;164216239;13503038;w?http://sf.net/marketplace

the external module (once patched) will only be able to build against 2.6.17 upwards because the in-kernel PIC/IOAPIC/LAPIC requires hrtimers and that functionality was only made public to external modules during the stabilization of 2.6.17 as seen by :

http://git.kernel.org/?p=linux/kernel/git/torvalds/linux-2.6.git;a=commitdiff;h=8d16b76421f0b3216012ee2d7819355e1cb847e5

there are obviously 3 possible fixes :

1) rewrite the in-kernel PIC/IOAPIC/LAPIC not to use hrtimers (not worthy)
2) replicate the functionality needed in the compatibility header or with the use of a helper (not that simple)
3) disable the functionality and ifdef the code for kernel < 2.6.17 (preferred)

but before, I would like to ask: is 2.6.9 upwards support really needed?, the only reason why I would suspect that 2.6.9 was recommended was because of RHEL4 but there is no kvm package available for it from Red Hat or in EPEL and Full Support is due May 15, 2008 so there might not be one ever, specially considering that Red Hat is focused on RHEL 5 Host for virtualization support as can be seen by :

http://www.redhat.com/rhel/virtualization/

other Linux distributions that are still under support will be affected as well (mainly because they are corporate linux focused and with longer supported timelines than community oriented distributions) but AFAIK they don't have yet a kvm package (and might never had one) :

* Ubuntu Dapper 6.06.1 LTS (2.6.15)
* SuSE Linux 10.1 (2.6.16)
* Mandriva Corporate 4 (2.6.12)

of course ALL community oriented distributions (fedora, ubuntu, gentoo) which are currently packaging kvm are ok and won't be affected by a "2.6.17 upwards" requirement.

Carlo

PS. shouldn't the 2.6.9 or higher requirement be in this page?
http://kvm.qumranet.com/kvmwiki/Downloads

-------------------------------------------------------------------------
SF.Net email is sponsored by:
Check out the new SourceForge.net Marketplace.
It's the best place to buy or sell services
for just about anything Open Source.
http://ad.doubleclick.net/clk;164216239;13503038;w?http://sf.net/marketplace

I imagine you're right, since there are few complaints. Most people have migrated to RHEL 5 or its equivalents.

Yes.

error compiling committee.c: too many arguments to function


-------------------------------------------------------------------------
SF.Net email is sponsored by:
Check out the new SourceForge.net Marketplace.
It's the best place to buy or sell services
for just about anything Open Source.
http://ad.doubleclick.net/clk;164216239;13503038;w?http://sf.net/marketplace