Back to Home / #uml / 2007 / 07 / Prev Day | Next Day
#uml IRC Logs for 2007-07-23

---Logopened Mon Jul 23 00:00:33 2007
00:51|-|balbir [~balbir@] has joined #uml
01:51|-|Ancalagon [] has joined #uml
02:18|-|xav2_ [] has quit [Remote host closed the connection]
02:19|-|xav2_ [~xav_@] has joined #uml
02:36|-|yairhr [] has quit [Quit: Leaving]
05:31|-|xav2_ [~xav_@] has quit [Ping timeout: 480 seconds]
06:03|-|kokoko1 [~Slacker@] has joined #uml
06:06|-|xav2_ [] has joined #uml
06:29|-|tyler [] has joined #uml
07:17|-|tyler [] has quit [Read error: Connection reset by peer]
07:19|-|tyler [] has joined #uml
07:59|-|tyler [] has quit [Read error: Connection reset by peer]
07:59|-|tyler [] has joined #uml
08:14|-|baroni [] has joined #uml
08:42|-|SixF9 [] has joined #uml
08:42|-|baroni [] has quit [Quit: Saindo]
09:21|-|dang [] has joined #uml
09:22|-|Ancalagon [] has quit [Ping timeout: 480 seconds]
09:22|-|Ancalagon [] has joined #uml
09:52|-|hfb [] has joined #uml
10:17|-|jdike [] has joined #uml
10:17<jdike>Hi guys
10:19|-|tyler [] has quit [Read error: Connection reset by peer]
10:19|-|xav2_ [] has quit [Quit: WeeChat 0.2.5]
10:20|-|tyler [] has joined #uml
10:24<kokoko1>hi jdike
10:40<dgraves>morning jdike!
10:42|-|tyler [] has quit [Ping timeout: 480 seconds]
10:46|-|tyler [] has joined #uml
10:52<kokoko1>so finaly xen made into mainline kernel
10:55<dang>Well, domU did, I think.
10:55<dang>But not dom0
10:56<kokoko1>hmm what is lguest ?
10:56<kokoko1>two more virtulizaiton into mainline xen and lguest
10:57<jdike>lguest is a simple, paravirt-ops-based guest
10:57<dang>lguest is kinda like UML, but virtualized in kernel space.
10:58<dang>Simple linux-only guest.
10:58<dang>The idea is to make it easy to use, and with as simple code as possible, so easy to understand.
10:58<kokoko1>ah nice
11:00<jdike>it's almost documentation
11:01<dang>And the simplest possible test case for paravirt-ops.
11:01<dang>And an argument for keeping paravirt-ops easy. :)
11:01<dang>er simple
11:01<jdike>an existence proof
11:05<kokoko1>look like lguest is targeted for kernel hackers not for production use ?
11:08<dang>Well, it's designed to be usable. So, you should be able to use it in similar circumstances to UML (if you're using UML to virtualize, and not, say, as a network simulator)
11:08<dang>I'm not sure it's ready for general use yet, tho.
11:10<kokoko1>too much virtualizations :), and look like everything is now revolving around kvm
11:11<dang>Well, the goal is to have a single virtualization framework in the kernel that all the virtualizers can use.
11:11<dang>So xen, vmware, uml, openvirt, and so on.
11:37<jdike>not really
11:37<jdike>the goal is provide many of the virtualizers a common framework
11:37<jdike>vmware, xen, and lguest can use pvops
11:38<jdike>the arch-based ones - UML, s390, ppc - can't
11:49[~]jdike starts beating on 2.6.23-rc1
12:01<jdike>... and in the meantime, I have some breakage to fix
12:06<jdike>... a bunch of missing __inits fixed
12:08|-|kos_tom [] has joined #uml
12:15<jdike>now on to the str* functions which are undefined all of a sudden
12:19<kokoko1>ahaha without any advance warning ? :)
12:20<jdike>since when do I get advance warning
12:22<dgraves>jdike: you really should fix that. next time, let them know they'll be fired if they dont' give you at least 2 weeks notice.
12:22<jdike>that's a plan
12:26<kokoko1>sorry i'm not very familiar with i tho they 'kernel folks' should be giving some advance warning when removing some function eg str*
12:26<dgraves>kokoko1: this is linux. the only folks that get any advance warning are the folks that write it.
12:26<dgraves>i doubt they got removed though.
12:26<jdike>they didn't
12:27<jdike>it looks like some restructuring of i386 which left UML out in the cold a little
12:28<jdike>I am stealing headers from i386 which say "this arch has its own implementations of the string functions", but not stealing the actual implementations
12:29<jdike>so I guess this is fixed with something like this
12:29<jdike>Index: linux-2.6.17/arch/um/sys-i386/Makefile
12:29<jdike>--- linux-2.6.17.orig/arch/um/sys-i386/Makefile 2007-07-23 13:29:01.000000000 -0400
12:29<jdike>+++ linux-2.6.17/arch/um/sys-i386/Makefile 2007-07-23 13:29:16.000000000 -0400
12:29<jdike>@@ -4,7 +4,7 @@ obj-y = bug.o bugs.o checksum.o delay.o
12:29<jdike> obj-$(CONFIG_MODE_SKAS) += stub.o stub_segv.o
12:30<jdike>-subarch-obj-y = lib/bitops.o lib/semaphore.o
12:30<jdike>+subarch-obj-y = lib/bitops.o lib/semaphore.o lib/string.o
12:30<jdike> subarch-obj-$(CONFIG_HIGHMEM) += mm/highmem.o
12:30<jdike> subarch-obj-$(CONFIG_MODULES) += kernel/module.o
12:30<jdike>but I haven't reached the end of the build to find out yet
12:30[~]kokoko1 ponder
12:33<dgraves>jdike: that's what you get for stealing headers....
12:33<jdike>a small price to pay to not have to reimplement or copy this stuff
12:34<dgraves>yeah, i know.
12:34<dgraves>it just always seems to bite you later. :(
12:46<jdike>WARNING: vmlinux: 'strlen' exported twice. Previous export was in vmlinux
12:46<jdike>I appear to have overdone this a bit
12:49<dgraves>jdike: better more than less, eh? :)
12:51<jdike>once for libc's strlen, once for the one I get from i386
12:55|-|richardw [] has joined #uml
12:56<dgraves>jdike: gotta love complications, eh?
12:56<jdike>no biggie
12:57<jdike>esp for a -rc1
13:04<dgraves>speaks the battle hardened warrior.
13:04[~]jdike admires his scars
13:07[~]balbir wonders if there is a fix he can find for 2.6.22-rc6-mm1 for it to compile (lib/check_signature.c) and boot
13:07<jdike>you can grab 2.6.23-rc2 which has this fixed
13:08<jdike>that mess got into 2.6.23 somehow, and was fixed in rc1
13:08<jdike>it broke a bunch of arches, UML included
13:09<jdike>something is still b0rked
13:09<jdike>WARNING: vmlinux(.bss+0x8204164): Section mismatch: reference to .init.setup:__setup_quiet_kernel (between 'stderr@@GLIBC_2.0' and 'completed.5757')
13:09<jdike>08204164 B stderr@@GLIBC_2.0
13:09<jdike>08204168 b completed.5757
13:09<jdike>between 'stderr@@GLIBC_2.0' and 'completed.5757' there are 0 bytes
13:10<jdike>and somewhere in those 0 bytes I have somehow stored a reference to a symbol in a non-__init section
13:21<balbir>jdike: thanks
13:22<balbir>jdike: I use -mm because of the container patches, I'll see if I can backport some of them to that version
13:24<balbir>jdike: I see 4 bytes b/w stderr@GLIBC_2.0 and completed.5757
13:24<balbir>jdike: from your output
13:30[~]balbir should stop commenting so late in the night :-)
13:33|-|tyler [] has quit [Ping timeout: 480 seconds]
13:55<jdike>balbir, there's a simpler patch, if you just want things to work agani
13:56<jdike>Index: linux-2.6.21-mm/include/asm-um/io.h
13:56<jdike>--- linux-2.6.21-mm.orig/include/asm-um/io.h 2007-02-04 13:44:54.000000000 -0
13:56<jdike>+++ linux-2.6.21-mm/include/asm-um/io.h 2007-06-28 12:05:24.000000000 -0400
13:56<jdike>@@ -33,6 +33,11 @@ static inline void * phys_to_virt(unsign
13:56<jdike> */
13:56<jdike> #define xlate_dev_kmem_ptr(p) p
13:56<jdike>+static inline unsigned char readb(const volatile void __iomem *addr)
13:56<jdike>+ return *(volatile unsigned char __force *) addr;
13:56<jdike> static inline void writeb(unsigned char b, volatile void __iomem *addr)
13:56<jdike> {
13:56<jdike> *(volatile unsigned char __force *) addr = b;
13:56<jdike>just add a readb implementation to io.h
13:57|-|baroni [] has joined #uml
14:33<jdike>a smallish batch of fixes off to Andrew
14:42|-|richardw_ [] has joined #uml
14:49|-|richardw [] has quit [Ping timeout: 480 seconds]
15:13|-|richardw_ [] has quit [Quit: Leaving]
15:16|-|horst [] has joined #uml
16:55|-|kos_tom [] has quit [Quit: I like core dumps]
16:59|-|baroni [] has quit [Ping timeout: 480 seconds]
17:27|-|dang [] has quit [Quit: Leaving.]
17:35|-|horst [] has quit [Remote host closed the connection]
17:58|-|hfb [] has quit [Quit: Leaving]
18:47|-|jdike [] has quit [Quit: Leaving]
22:59|-|VS_ChanLog [] has left #uml [Rotating Logs]
22:59|-|VS_ChanLog [] has joined #uml
23:54|-|balbir [~balbir@] has quit [Ping timeout: 480 seconds]
---Logclosed Tue Jul 24 00:00:07 2007