Boot Problem (and fix)

Boot Problem (and fix)

[the uni]

Starting with version 77 (I believe) I can no longer boot any of my
virtual machines. Just hangs at "press f12 for boot menu".

My specs: core2duo, gentoo 2008 stable/current,x86_64, 6gigs ram.

I managed to solve the problem by replacing the bios.bin in
/usr/share/kvm/ with an older one. It seems to have been broken by:

commit 0ad566e0d8780b465f8af17926ec7b73418ef076
date: 09/23/2008

I've seen a few other people with the same issue. Hope this helps someone

TheUni

Re: Boot Problem (and fix)

[Avi Kivity]

the uni wrote:
> Starting with version 77 (I believe) I can no longer boot any of my
> virtual machines. Just hangs at "press f12 for boot menu".
>
> My specs: core2duo, gentoo 2008 stable/current,x86_64, 6gigs ram.
>
> I managed to solve the problem by replacing the bios.bin in
> /usr/share/kvm/ with an older one. It seems to have been broken by:
>
> commit 0ad566e0d8780b465f8af17926ec7b73418ef076
> date: 09/23/2008
>
> I've seen a few other people with the same issue. Hope this helps someone
>
>   

What are the parameters you use to launch your guests (memory size, smp)?

-- 
error compiling committee.c: too many arguments to function

Re: Boot Problem (and fix)

[the uni]

   > What are the parameters you use to launch your guests (memory size, smp)?

Nothing special. kvm disk.img is enough to freeze the VM. Memory, smp,
etc can all be changed with no effect.

The only parameter that allows me to boot is -no-kvm

As I stated before, everything is normal with an older bios image.

Same issue seen here:
http://kerneltrap.org/mailarchive/linux-kvm/2008/11/2/3900354

Re: Boot Problem (and fix)

[Avi Kivity]

the uni wrote:
>    > What are the parameters you use to launch your guests (memory size, smp)?
>
> Nothing special. kvm disk.img is enough to freeze the VM. Memory, smp,
> etc can all be changed with no effect.
>
> The only parameter that allows me to boot is -no-kvm
>
> As I stated before, everything is normal with an older bios image.
>
> Same issue seen here:
> http://kerneltrap.org/mailarchive/linux-kvm/2008/11/2/3900354
>   

Please post what 'info registers' and 'x/40b $eip' produce on the 
monitor ('-monitor stdio' helps).

-- 
I have a truly marvellous patch that fixes the bug which this
signature is too narrow to contain.

Re: Boot Problem (and fix)

[the uni]

> Please post what 'info registers' and 'x/40b $eip' produce on the monitor
> ('-monitor stdio' helps).

while hung on non-functioning bios:

(gdb) info registers
rax            0xfffffffffffffdfe       -514
rbx            0xe      14
rcx            0xffffffffffffffff       -1
rdx            0x7fffa4c81b70   140735957965680
rsi            0x7fffa4c81bf0   140735957965808
rdi            0xf      15
rbp            0x1      0x1
rsp            0x7fffa4c81ab0   0x7fffa4c81ab0
r8             0x7fffa4c81c70   140735957965936
r9             0xb40    2880
r10            0x7fffa4c81af0   140735957965552
r11            0x202    514
r12            0x7fffa4c81af0   140735957965552
r13            0x7fffa4c81b70   140735957965680
r14            0x7fffa4c81bf0   140735957965808
r15            0x1      1
rip            0x7f329b5ad0c2   0x7f329b5ad0c2 <select+98>
eflags         0x202    [ IF ]
cs             0x33     51
ss             0x2b     43
ds             0x0      0
es             0x0      0
fs             0x0      0
gs             0x0      0
fctrl          0x37f    895
fstat          0x0      0
ftag           0xffff   65535
fiseg          0x0      0
fioff          0x0      0
foseg          0x0      0
fooff          0x0      0
fop            0x0      0
mxcsr          0x1f80   [ IM DM ZM OM UM PM ]


(gdb) x/40b $eip
Value can't be converted to integer.


(gdb) x/40b $rip
0x7f329b5ad0c2 <select+98>:     72 'H'  -117 '\213'     60 '<'  36 '$'
 72 'H' -119 '\211'      68 'D'  36 '$'
0x7f329b5ad0ca <select+106>:    8 '\b'  -24 'è' -112 '\220'     16
'\020'      1 '\001' 0 '\0'  72 'H'  -117 '\213'
0x7f329b5ad0d2 <select+114>:    68 'D'  36 '$'  8 '\b'  72 'H'  -125
'\203'    -60 'Ä'  56 '8'  72 'H'
0x7f329b5ad0da <select+122>:    61 '='  1 '\001'        -16 'ð' -1 'ÿ'
 -1 'ÿ' 115 's'  1 '\001'        -61 'Ã'
0x7f329b5ad0e2 <select+130>:    72 'H'  -117 '\213'     13 '\r' -65
'¿' -18 'î'39 '\''  0 '\0'  49 '1'

(gdb) x/i $rip
0x7f329b5ad0c2 <select+98>:     mov    (%rsp),%rdi

Re: Boot Problem (and fix)

[the uni]

Sorry, just realized what you meant

(qemu) info registers
EAX=0000f3b3 EBX=00000012 ECX=0000e080 EDX=00000012
ESI=ffff0000 EDI=0008fdba EBP=0000ffb0 ESP=0000ff9e
EIP=000008d2 EFL=00033202 [-------] CPL=3 II=0 A20=1 SMM=0 HLT=0
ES =0080 00000800 0000ffff 0000f300
CS =f000 000f0000 0000ffff 0000f300
SS =0000 00000000 0000ffff 0000f300
DS =0000 00000000 0000ffff 0000f300
FS =0000 00000000 0000ffff 0000f300
GS =0000 00000000 0000ffff 0000f300
LDT=0000 00000000 0000ffff 00008200
TR =0000 fffbd000 00002088 00008b00
GDT=     000fb812 00000030
IDT=     00000000 000003ff
CR0=00000010 CR2=00000000 CR3=00000000 CR4=00000000
FCW=037f FSW=0000 [ST=0] FTW=00 MXCSR=00000000
FPR0=0000000000000000 0000 FPR1=0000000000000000 0000
FPR2=0000000000000000 0000 FPR3=0000000000000000 0000
FPR4=0000000000000000 0000 FPR5=0000000000000000 0000
FPR6=0000000000000000 0000 FPR7=0000000000000000 0000
XMM00=00000000000000000000000000000000 XMM01=00000000000000000000000000000000
XMM02=00000000000000000000000000000000 XMM03=00000000000000000000000000000000
XMM04=00000000000000000000000000000000 XMM05=00000000000000000000000000000000
XMM06=00000000000000000000000000000000 XMM07=00000000000000000000000000000000


(qemu) x/40b $eip
00000000000008d2: 0x00 0x00 0x00 0x00 0x00 0x00 0x00 0x00
00000000000008da: 0x00 0x00 0x00 0x00 0x00 0x00 0x00 0x00
00000000000008e2: 0x00 0x00 0x00 0x00 0x00 0x00 0x00 0x00
00000000000008ea: 0x00 0x00 0x00 0x00 0x00 0x00 0x00 0x00
00000000000008f2: 0x00 0x00 0x00 0x00 0x00 0x00 0x00 0x00

(qemu) x/i $eip
0x00000000000008d2:  add    %al,(%bx,%si)

Re: Boot Problem (and fix)

[Avi Kivity]

the uni wrote:
> Sorry, just realized what you meant
>
> (qemu) info registers
> EAX=0000f3b3 EBX=00000012 ECX=0000e080 EDX=00000012
> ESI=ffff0000 EDI=0008fdba EBP=0000ffb0 ESP=0000ff9e
> EIP=000008d2 EFL=00033202 [-------] CPL=3 II=0 A20=1 SMM=0 HLT=0
> ES =0080 00000800 0000ffff 0000f300
> CS =f000 000f0000 0000ffff 0000f300
> SS =0000 00000000 0000ffff 0000f300
> DS =0000 00000000 0000ffff 0000f300
> FS =0000 00000000 0000ffff 0000f300
> GS =0000 00000000 0000ffff 0000f300
> LDT=0000 00000000 0000ffff 00008200
> TR =0000 fffbd000 00002088 00008b00
> GDT=     000fb812 00000030
> IDT=     00000000 000003ff
> CR0=00000010 CR2=00000000 CR3=00000000 CR4=00000000
> FCW=037f FSW=0000 [ST=0] FTW=00 MXCSR=00000000
> FPR0=0000000000000000 0000 FPR1=0000000000000000 0000
> FPR2=0000000000000000 0000 FPR3=0000000000000000 0000
> FPR4=0000000000000000 0000 FPR5=0000000000000000 0000
> FPR6=0000000000000000 0000 FPR7=0000000000000000 0000
> XMM00=00000000000000000000000000000000 XMM01=00000000000000000000000000000000
> XMM02=00000000000000000000000000000000 XMM03=00000000000000000000000000000000
> XMM04=00000000000000000000000000000000 XMM05=00000000000000000000000000000000
> XMM06=00000000000000000000000000000000 XMM07=00000000000000000000000000000000
>
>   

With which bios was this taken?  Please supply sha1sum of bios.bin.

> (qemu) x/40b $eip
> 00000000000008d2: 0x00 0x00 0x00 0x00 0x00 0x00 0x00 0x00
> 00000000000008da: 0x00 0x00 0x00 0x00 0x00 0x00 0x00 0x00
> 00000000000008e2: 0x00 0x00 0x00 0x00 0x00 0x00 0x00 0x00
> 00000000000008ea: 0x00 0x00 0x00 0x00 0x00 0x00 0x00 0x00
> 00000000000008f2: 0x00 0x00 0x00 0x00 0x00 0x00 0x00 0x00
>
> (qemu) x/i $eip
> 0x00000000000008d2:  add    %al,(%bx,%si)
>   

My mistake: $eip + 0xf0000 would be more informative.

-- 
error compiling committee.c: too many arguments to function

Re: Boot Problem (and fix)

[the uni]

>My mistake: $eip + 0xf0000 would be more informative.

(qemu) x/40b $eip + 0xf0000
00000000000f08d2: 0xf4 0xb8 0x6c 0x04 0x50 0x31 0xc0 0x50
00000000000f08da: 0xe8 0xa6 0xf7 0x89 0xd3 0x83 0xc4 0x04
00000000000f08e2: 0x89 0x46 0xf0 0x89 0x5e 0xf2 0x8b 0x46
00000000000f08ea: 0xf4 0x8b 0x5e 0xf6 0x8d 0x7e 0xf0 0xe8
00000000000f08f2: 0xd7 0xf7 0x73 0x26 0x8b 0x46 0xf0 0x8b

>With which bios was this taken?  Please supply sha1sum of bios.bin.

ae04986718470aeca71325992a54204794de81d5  /usr/share/kvm/bios.bin

Re: Boot Problem (and fix)

[Avi Kivity]

the uni wrote:
>> My mistake: $eip + 0xf0000 would be more informative.
>>     
>
> (qemu) x/40b $eip + 0xf0000
> 00000000000f08d2: 0xf4 0xb8 0x6c 0x04 0x50 0x31 0xc0 0x50
> 00000000000f08da: 0xe8 0xa6 0xf7 0x89 0xd3 0x83 0xc4 0x04
> 00000000000f08e2: 0x89 0x46 0xf0 0x89 0x5e 0xf2 0x8b 0x46
> 00000000000f08ea: 0xf4 0x8b 0x5e 0xf6 0x8d 0x7e 0xf0 0xe8
> 00000000000f08f2: 0xd7 0xf7 0x73 0x26 0x8b 0x46 0xf0 0x8b
>
>   

The cpu is blocked on a hlt instruction, waiting for a keypress from an 
interrupt.

What is your host kernel?  And are the kvm-77 modules loaded, or are you 
running the modules that came with that kernel?

(If running the kvm-77 modules, there should be a message in dmesg to 
that effect).

-- 
error compiling committee.c: too many arguments to function

Re: Boot Problem (and fix)

[the uni]

> What is your host kernel?  And are the kvm-77 modules loaded, or are you
> running the modules that came with that kernel?

It's happening to me on both of my machines:

Linux mybox 2.6.25-gentoo-r7 #12 SMP PREEMPT Sat Nov 8 01:16:45 EST 2008 x86_64
and
Linux server2900 2.6.26-gentoo #4 SMP Sat Sep 20 16:10:17 EDT 2008 x86_64

On both, I'm using the modules that came with the kernel -- compiled in.

Re: Boot Problem (and fix)

[Avi Kivity]

the uni wrote:
>> What is your host kernel?  And are the kvm-77 modules loaded, or are you
>> running the modules that came with that kernel?
>>     
>
> It's happening to me on both of my machines:
>
> Linux mybox 2.6.25-gentoo-r7 #12 SMP PREEMPT Sat Nov 8 01:16:45 EST 2008 x86_64
> and
> Linux server2900 2.6.26-gentoo #4 SMP Sat Sep 20 16:10:17 EDT 2008 x86_64
>
> On both, I'm using the modules that came with the kernel -- compiled in.
>   

Can you try the modules that came with kvm-77, just to verify that's 
it's a kernel problem that's been fixed?  If so, we can find it and 
backport the fix.

-- 
error compiling committee.c: too many arguments to function

Re: Boot Problem (and fix)

[the uni]

Confirmed that vers 77 and 78 both work fine.

Re: Boot Problem (and fix)

[Avi Kivity]

the uni wrote:
> Confirmed that vers 77 and 78 both work fine.
>   

Can you try modules from older versions (keeping userspace and the bios) 
to see which version introduced the fix?

-- 
error compiling committee.c: too many arguments to function

Re: Boot Problem (and fix)

[the uni]

> Can you try modules from older versions (keeping userspace and the bios) to
> see which version introduced the fix?


Keeping userspace and bios, versions 71 and up are fixed... 70 is broken.

Re: Boot Problem (and fix)

[Avi Kivity]

[-- Attachment #1: Type: text/plain, Size: 396 bytes --]

the uni wrote:
>> Can you try modules from older versions (keeping userspace and the bios) to
>> see which version introduced the fix?
>>     
>
>
> Keeping userspace and bios, versions 71 and up are fixed... 70 is broken.
>   

Please try kvm-70 with the attached patch attached.

(from 214fc251236fd1d0e0f2b7c3b7b7ad5243928fa9)

-- 
error compiling committee.c: too many arguments to function


[-- Attachment #2: fix-halt.patch --]
[-- Type: text/plain, Size: 1043 bytes --]

commit 214fc251236fd1d0e0f2b7c3b7b7ad5243928fa9
Author: Mohammed Gamal <m.gamal005@gmail.com>
Date:   Sun Jul 6 16:51:26 2008 +0300

    KVM: x86 emulator: Fix HLT instruction
    
    This patch fixes issue encountered with HLT instruction
    under FreeDOS's HIMEM XMS Driver.
    
    The HLT instruction jumped directly to the done label and
    skips updating the EIP value, therefore causing the guest
    to spin endlessly on the same instruction.
    
    The patch changes the instruction so that it writes back
    the updated EIP value.
    
    Signed-off-by: Mohammed Gamal <m.gamal005@gmail.com>
    Signed-off-by: Avi Kivity <avi@qumranet.com>

diff --git a/kernel/x86_emulate.c b/kernel/x86_emulate.c
index dd4efe1..62e71b6 100644
--- a/kernel/x86_emulate.c
+++ b/kernel/x86_emulate.c
@@ -1732,7 +1732,7 @@ special_insn:
 		break;
 	case 0xf4:              /* hlt */
 		ctxt->vcpu->arch.halt_request = 1;
-		goto done;
+		break;
 	case 0xf5:	/* cmc */
 		/* complement carry flag from eflags reg */
 		ctxt->eflags ^= EFLG_CF;

Re: Boot Problem (and fix)

[the uni]

That fixed it. Thanks.

Re: Boot Problem (and fix)

[Avi Kivity]

the uni wrote:
> That fixed it. Thanks.
>   

Ok.  Since 2.6.25 is out of maintenance mode I suggest you forward this 
patch to your distro's kernel maintainers.  I'll queue it for 2.6.2[67].

-- 
error compiling committee.c: too many arguments to function

Re: Boot Problem (and fix)

[Avi Kivity]

Avi Kivity wrote:
> Ok.  Since 2.6.25 is out of maintenance mode I suggest you forward 
> this patch to your distro's kernel maintainers.  I'll queue it for 
> 2.6.2[67].
>

2.6.26.1 has the fix, so all you need to do is upgrade to that or a 
later version.

-- 
I have a truly marvellous patch that fixes the bug which this
signature is too narrow to contain.

Re: Boot Problem (and fix)

[the uni]

> 2.6.26.1 has the fix, so all you need to do is upgrade to that or a later
> version.

OK. Gentoo stable is still at 2.6.25, I'll report it there.