* KVM emulation failure with "movups" instruction
@ 2017-08-04 22:03 Anatol Pomozov
2017-08-04 22:41 ` Wanpeng Li
2017-08-07 16:59 ` David Matlack
0 siblings, 2 replies; 7+ messages in thread
From: Anatol Pomozov @ 2017-08-04 22:03 UTC (permalink / raw)
To: kvm; +Cc: Paolo Bonzini, Dan Carpenter, david, dmatlack
Hello
I am testing my own x86_64 kernel with QEMU/KVM and it report cryptic
error "KVM internal error. Suberror: 1". I believe this emulation
error happens in Linux kernel most likely in arch/x86/kvm/emulate.c
I tested KVM with Linux 4.4.0, 4.12.3 and both versions return this
emulation error. It is strange as QEMU emulation and VMWare work fine
with the same kernel binary.
How to understand what is the KVM issue and why it fails? Is there a
way to obtain more information about the error?
Here is my code where the issue happens:
0000000000100c20 <console_write>:
100c20: 48 85 f6 test %rsi,%rsi
100c23: 0f 84 11 05 00 00 je 10113a <console_write+0x51a>
100c29: 45 31 c0 xor %r8d,%r8d
100c2c: b9 0e 8f 0b 00 mov $0xb8f0e,%ecx
100c31: 66 66 66 66 66 66 2e data32 data32 data32 data32
data32 nopw %cs:0x0(%rax,%rax,1)
100c38: 0f 1f 84 00 00 00 00
100c3f: 00
100c40: 0f be 07 movsbl (%rdi),%eax
100c43: 0f b7 d0 movzwl %ax,%edx
100c46: 83 fa 0a cmp $0xa,%edx
100c49: 0f 85 91 00 00 00 jne 100ce0 <console_write+0xc0>
100c4f: 48 8b 05 aa 83 00 00 mov 0x83aa(%rip),%rax
# 109000 <row>
100c56: 48 8d 50 01 lea 0x1(%rax),%rdx
100c5a: 48 89 15 9f 83 00 00 mov %rdx,0x839f(%rip)
# 109000 <row>
100c61: 48 83 fa 19 cmp $0x19,%rdx
100c65: 0f 85 35 04 00 00 jne 1010a0 <console_write+0x480>
100c6b: 31 d2 xor %edx,%edx
100c6d: 0f 1f 00 nopl (%rax)
100c70: 0f 10 84 51 92 f1 ff movups -0xe6e(%rcx,%rdx,2),%xmm0
^^^^^^^^^^^^ KVM error happens here ^^^^^^^ See registers stack below
100c77: ff
100c78: 0f 11 84 51 f2 f0 ff movups %xmm0,-0xf0e(%rcx,%rdx,2)
100c7f: ff
100c80: 48 83 c2 08 add $0x8,%rdx
100c84: 48 81 fa 80 07 00 00 cmp $0x780,%rdx
100c8b: 75 e3 jne 100c70 <console_write+0x50>
100c8d: 48 89 05 6c 83 00 00 mov %rax,0x836c(%rip)
# 109000 <row>
100c94: 48 c7 05 c1 3b 00 00 movq $0xb8f00,0x3bc1(%rip)
# 104860 <cursor>
100c9b: 00 8f 0b 00
100c9f: 4c 8b 0d 62 83 00 00 mov 0x8362(%rip),%r9
# 109008 <column>
100ca6: 4d 85 c9 test %r9,%r9
100ca9: 0f 84 71 04 00 00 je 101120 <console_write+0x500>
100caf: 49 8d 41 ff lea -0x1(%r9),%rax
100cb3: 44 89 ca mov %r9d,%edx
100cb6: 83 e2 07 and $0x7,%edx
100cb9: 48 83 f8 07 cmp $0x7,%rax
100cbd: 0f 83 f6 03 00 00 jae 1010b9 <console_write+0x499>
Registers at KVM error as reported by QEMU
RAX=0000000000000018 RBX=0000000000000001 RCX=00000000000b8f0e
RDX=0000000000000000
RSI=0000000000000001 RDI=000000000010445a RBP=000000000010445b
RSP=000000000010af78
R8 =0000000000000000 R9 =00000000000b8f56 R10=000000000010445c
R11=0000000000000004
R12=000000000010445a R13=000000007fffffd3 R14=0000000000000001
R15=000000000000002c
RIP=0000000000100c70 RFL=00010246 [---Z-P-] CPL=0 II=0 A20=1 SMM=0 HLT=0
ES =0000 0000000000000000 ffffffff 00c00000
CS =0008 0000000000000000 00000000 00209900 DPL=0 CS64 [--A]
SS =0000 0000000000000000 ffffffff 00c00000
DS =0000 0000000000000000 ffffffff 00c00000
FS =0000 0000000000000000 ffffffff 00c00000
GS =0000 0000000000000000 ffffffff 00c00000
LDT=0000 0000000000000000 0000ffff 00008200 DPL=0 LDT
TR =0000 0000000000000000 0000ffff 00008b00 DPL=0 TSS64-busy
GDT= 0000000000104518 0000000f
IDT= 0000000000105030 00000fff
CR0=80000033 CR2=0000000000000000 CR3=0000000000107000 CR4=00000620
DR0=0000000000000000 DR1=0000000000000000 DR2=0000000000000000
DR3=0000000000000000
DR6=00000000ffff0ff0 DR7=0000000000000400
EFER=0000000000000500
^ permalink raw reply [flat|nested] 7+ messages in thread
* Re: KVM emulation failure with "movups" instruction
2017-08-04 22:03 KVM emulation failure with "movups" instruction Anatol Pomozov
@ 2017-08-04 22:41 ` Wanpeng Li
2017-08-07 16:59 ` David Matlack
1 sibling, 0 replies; 7+ messages in thread
From: Wanpeng Li @ 2017-08-04 22:41 UTC (permalink / raw)
To: Anatol Pomozov
Cc: kvm, Paolo Bonzini, Dan Carpenter, David Hildenbrand, David Matlack
2017-08-05 6:03 GMT+08:00 Anatol Pomozov <anatol.pomozov@gmail.com>:
> Hello
>
> I am testing my own x86_64 kernel with QEMU/KVM and it report cryptic
> error "KVM internal error. Suberror: 1". I believe this emulation
> error happens in Linux kernel most likely in arch/x86/kvm/emulate.c
>
> I tested KVM with Linux 4.4.0, 4.12.3 and both versions return this
> emulation error. It is strange as QEMU emulation and VMWare work fine
> with the same kernel binary.
>
>
> How to understand what is the KVM issue and why it fails? Is there a
> way to obtain more information about the error?
Coud you post your testcase?
Regards,
Wanpeng Li
>
>
> Here is my code where the issue happens:
>
> 0000000000100c20 <console_write>:
> 100c20: 48 85 f6 test %rsi,%rsi
> 100c23: 0f 84 11 05 00 00 je 10113a <console_write+0x51a>
> 100c29: 45 31 c0 xor %r8d,%r8d
> 100c2c: b9 0e 8f 0b 00 mov $0xb8f0e,%ecx
> 100c31: 66 66 66 66 66 66 2e data32 data32 data32 data32
> data32 nopw %cs:0x0(%rax,%rax,1)
> 100c38: 0f 1f 84 00 00 00 00
> 100c3f: 00
> 100c40: 0f be 07 movsbl (%rdi),%eax
> 100c43: 0f b7 d0 movzwl %ax,%edx
> 100c46: 83 fa 0a cmp $0xa,%edx
> 100c49: 0f 85 91 00 00 00 jne 100ce0 <console_write+0xc0>
> 100c4f: 48 8b 05 aa 83 00 00 mov 0x83aa(%rip),%rax
> # 109000 <row>
> 100c56: 48 8d 50 01 lea 0x1(%rax),%rdx
> 100c5a: 48 89 15 9f 83 00 00 mov %rdx,0x839f(%rip)
> # 109000 <row>
> 100c61: 48 83 fa 19 cmp $0x19,%rdx
> 100c65: 0f 85 35 04 00 00 jne 1010a0 <console_write+0x480>
> 100c6b: 31 d2 xor %edx,%edx
> 100c6d: 0f 1f 00 nopl (%rax)
> 100c70: 0f 10 84 51 92 f1 ff movups -0xe6e(%rcx,%rdx,2),%xmm0
>
> ^^^^^^^^^^^^ KVM error happens here ^^^^^^^ See registers stack below
>
> 100c77: ff
> 100c78: 0f 11 84 51 f2 f0 ff movups %xmm0,-0xf0e(%rcx,%rdx,2)
> 100c7f: ff
> 100c80: 48 83 c2 08 add $0x8,%rdx
> 100c84: 48 81 fa 80 07 00 00 cmp $0x780,%rdx
> 100c8b: 75 e3 jne 100c70 <console_write+0x50>
> 100c8d: 48 89 05 6c 83 00 00 mov %rax,0x836c(%rip)
> # 109000 <row>
> 100c94: 48 c7 05 c1 3b 00 00 movq $0xb8f00,0x3bc1(%rip)
> # 104860 <cursor>
> 100c9b: 00 8f 0b 00
> 100c9f: 4c 8b 0d 62 83 00 00 mov 0x8362(%rip),%r9
> # 109008 <column>
> 100ca6: 4d 85 c9 test %r9,%r9
> 100ca9: 0f 84 71 04 00 00 je 101120 <console_write+0x500>
> 100caf: 49 8d 41 ff lea -0x1(%r9),%rax
> 100cb3: 44 89 ca mov %r9d,%edx
> 100cb6: 83 e2 07 and $0x7,%edx
> 100cb9: 48 83 f8 07 cmp $0x7,%rax
> 100cbd: 0f 83 f6 03 00 00 jae 1010b9 <console_write+0x499>
>
>
>
>
>
> Registers at KVM error as reported by QEMU
>
> RAX=0000000000000018 RBX=0000000000000001 RCX=00000000000b8f0e
> RDX=0000000000000000
> RSI=0000000000000001 RDI=000000000010445a RBP=000000000010445b
> RSP=000000000010af78
> R8 =0000000000000000 R9 =00000000000b8f56 R10=000000000010445c
> R11=0000000000000004
> R12=000000000010445a R13=000000007fffffd3 R14=0000000000000001
> R15=000000000000002c
> RIP=0000000000100c70 RFL=00010246 [---Z-P-] CPL=0 II=0 A20=1 SMM=0 HLT=0
> ES =0000 0000000000000000 ffffffff 00c00000
> CS =0008 0000000000000000 00000000 00209900 DPL=0 CS64 [--A]
> SS =0000 0000000000000000 ffffffff 00c00000
> DS =0000 0000000000000000 ffffffff 00c00000
> FS =0000 0000000000000000 ffffffff 00c00000
> GS =0000 0000000000000000 ffffffff 00c00000
> LDT=0000 0000000000000000 0000ffff 00008200 DPL=0 LDT
> TR =0000 0000000000000000 0000ffff 00008b00 DPL=0 TSS64-busy
> GDT= 0000000000104518 0000000f
> IDT= 0000000000105030 00000fff
> CR0=80000033 CR2=0000000000000000 CR3=0000000000107000 CR4=00000620
> DR0=0000000000000000 DR1=0000000000000000 DR2=0000000000000000
> DR3=0000000000000000
> DR6=00000000ffff0ff0 DR7=0000000000000400
> EFER=0000000000000500
^ permalink raw reply [flat|nested] 7+ messages in thread
* Re: KVM emulation failure with "movups" instruction
2017-08-04 22:03 KVM emulation failure with "movups" instruction Anatol Pomozov
2017-08-04 22:41 ` Wanpeng Li
@ 2017-08-07 16:59 ` David Matlack
2017-08-07 18:45 ` Anatol Pomozov
1 sibling, 1 reply; 7+ messages in thread
From: David Matlack @ 2017-08-07 16:59 UTC (permalink / raw)
To: Anatol Pomozov; +Cc: kvm list, Paolo Bonzini, Dan Carpenter, David Hildenbrand
On Fri, Aug 4, 2017 at 3:03 PM, Anatol Pomozov <anatol.pomozov@gmail.com> wrote:
> Hello
>
> I am testing my own x86_64 kernel with QEMU/KVM and it report cryptic
> error "KVM internal error. Suberror: 1". I believe this emulation
> error happens in Linux kernel most likely in arch/x86/kvm/emulate.c
>
> I tested KVM with Linux 4.4.0, 4.12.3 and both versions return this
> emulation error. It is strange as QEMU emulation and VMWare work fine
> with the same kernel binary.
>
>
> How to understand what is the KVM issue and why it fails? Is there a
> way to obtain more information about the error?
>
>
> Here is my code where the issue happens:
>
> 0000000000100c20 <console_write>:
> 100c20: 48 85 f6 test %rsi,%rsi
> 100c23: 0f 84 11 05 00 00 je 10113a <console_write+0x51a>
> 100c29: 45 31 c0 xor %r8d,%r8d
> 100c2c: b9 0e 8f 0b 00 mov $0xb8f0e,%ecx
> 100c31: 66 66 66 66 66 66 2e data32 data32 data32 data32
> data32 nopw %cs:0x0(%rax,%rax,1)
> 100c38: 0f 1f 84 00 00 00 00
> 100c3f: 00
> 100c40: 0f be 07 movsbl (%rdi),%eax
> 100c43: 0f b7 d0 movzwl %ax,%edx
> 100c46: 83 fa 0a cmp $0xa,%edx
> 100c49: 0f 85 91 00 00 00 jne 100ce0 <console_write+0xc0>
> 100c4f: 48 8b 05 aa 83 00 00 mov 0x83aa(%rip),%rax
> # 109000 <row>
> 100c56: 48 8d 50 01 lea 0x1(%rax),%rdx
> 100c5a: 48 89 15 9f 83 00 00 mov %rdx,0x839f(%rip)
> # 109000 <row>
> 100c61: 48 83 fa 19 cmp $0x19,%rdx
> 100c65: 0f 85 35 04 00 00 jne 1010a0 <console_write+0x480>
> 100c6b: 31 d2 xor %edx,%edx
> 100c6d: 0f 1f 00 nopl (%rax)
> 100c70: 0f 10 84 51 92 f1 ff movups -0xe6e(%rcx,%rdx,2),%xmm0
Likely this instruction is accessing an MMIO region (probably a VGA
buffer, given "mov $0xb8f0e,%ecx"). This will cause a trap to KVM
and KVM will try to emulate the instruction. The instruction doing the
MMIO access is an SSE instruction, which KVM does not have support to
emulate, thus the internal error. If you didn't hand write this
assembly I would try compiling your kernel with compilation flags to
disable SSE code generation and see if that fixes your issue.
>
> ^^^^^^^^^^^^ KVM error happens here ^^^^^^^ See registers stack below
>
> 100c77: ff
> 100c78: 0f 11 84 51 f2 f0 ff movups %xmm0,-0xf0e(%rcx,%rdx,2)
> 100c7f: ff
> 100c80: 48 83 c2 08 add $0x8,%rdx
> 100c84: 48 81 fa 80 07 00 00 cmp $0x780,%rdx
> 100c8b: 75 e3 jne 100c70 <console_write+0x50>
> 100c8d: 48 89 05 6c 83 00 00 mov %rax,0x836c(%rip)
> # 109000 <row>
> 100c94: 48 c7 05 c1 3b 00 00 movq $0xb8f00,0x3bc1(%rip)
> # 104860 <cursor>
> 100c9b: 00 8f 0b 00
> 100c9f: 4c 8b 0d 62 83 00 00 mov 0x8362(%rip),%r9
> # 109008 <column>
> 100ca6: 4d 85 c9 test %r9,%r9
> 100ca9: 0f 84 71 04 00 00 je 101120 <console_write+0x500>
> 100caf: 49 8d 41 ff lea -0x1(%r9),%rax
> 100cb3: 44 89 ca mov %r9d,%edx
> 100cb6: 83 e2 07 and $0x7,%edx
> 100cb9: 48 83 f8 07 cmp $0x7,%rax
> 100cbd: 0f 83 f6 03 00 00 jae 1010b9 <console_write+0x499>
>
>
>
>
>
> Registers at KVM error as reported by QEMU
>
> RAX=0000000000000018 RBX=0000000000000001 RCX=00000000000b8f0e
> RDX=0000000000000000
> RSI=0000000000000001 RDI=000000000010445a RBP=000000000010445b
> RSP=000000000010af78
> R8 =0000000000000000 R9 =00000000000b8f56 R10=000000000010445c
> R11=0000000000000004
> R12=000000000010445a R13=000000007fffffd3 R14=0000000000000001
> R15=000000000000002c
> RIP=0000000000100c70 RFL=00010246 [---Z-P-] CPL=0 II=0 A20=1 SMM=0 HLT=0
> ES =0000 0000000000000000 ffffffff 00c00000
> CS =0008 0000000000000000 00000000 00209900 DPL=0 CS64 [--A]
> SS =0000 0000000000000000 ffffffff 00c00000
> DS =0000 0000000000000000 ffffffff 00c00000
> FS =0000 0000000000000000 ffffffff 00c00000
> GS =0000 0000000000000000 ffffffff 00c00000
> LDT=0000 0000000000000000 0000ffff 00008200 DPL=0 LDT
> TR =0000 0000000000000000 0000ffff 00008b00 DPL=0 TSS64-busy
> GDT= 0000000000104518 0000000f
> IDT= 0000000000105030 00000fff
> CR0=80000033 CR2=0000000000000000 CR3=0000000000107000 CR4=00000620
> DR0=0000000000000000 DR1=0000000000000000 DR2=0000000000000000
> DR3=0000000000000000
> DR6=00000000ffff0ff0 DR7=0000000000000400
> EFER=0000000000000500
^ permalink raw reply [flat|nested] 7+ messages in thread
* Re: KVM emulation failure with "movups" instruction
2017-08-07 16:59 ` David Matlack
@ 2017-08-07 18:45 ` Anatol Pomozov
2017-08-07 22:00 ` Wanpeng Li
0 siblings, 1 reply; 7+ messages in thread
From: Anatol Pomozov @ 2017-08-07 18:45 UTC (permalink / raw)
To: David Matlack; +Cc: kvm list, Paolo Bonzini, Dan Carpenter, David Hildenbrand
Hi David
On Mon, Aug 7, 2017 at 9:59 AM, David Matlack <dmatlack@google.com> wrote:
> Likely this instruction is accessing an MMIO region (probably a VGA
> buffer, given "mov $0xb8f0e,%ecx"). This will cause a trap to KVM
> and KVM will try to emulate the instruction. The instruction doing the
> MMIO access is an SSE instruction,
It is correct. The emulation error happens in printf() kernel function
that formats a string and then prints it VGA area. Compilers (both
clang and gcc) think that SSE instructions are more efficient in
dealing with strings and that's why I have SSE instructions here.
> which KVM does not have support to
> emulate, thus the internal error. If you didn't hand write this
> assembly I would try compiling your kernel with compilation flags to
> disable SSE code generation and see if that fixes your issue.
Adding "-mno-sse" and removing some floating point functionality in my
kernel make the emulation error go away. So it confirms your thesis.
Though I need floating point/SSE in my kernel and I can't disable SSE
completely. Instead I would love to understand why KVM can't use SSE
with VGA memory? I tested my kernel binary with bare hardware, KVM,
VMWare, QEMU and only KVM fails.
^ permalink raw reply [flat|nested] 7+ messages in thread
* Re: KVM emulation failure with "movups" instruction
2017-08-07 18:45 ` Anatol Pomozov
@ 2017-08-07 22:00 ` Wanpeng Li
2017-08-09 22:02 ` Anatol Pomozov
0 siblings, 1 reply; 7+ messages in thread
From: Wanpeng Li @ 2017-08-07 22:00 UTC (permalink / raw)
To: Anatol Pomozov
Cc: David Matlack, kvm list, Paolo Bonzini, Dan Carpenter, David Hildenbrand
2017-08-08 2:45 GMT+08:00 Anatol Pomozov <anatol.pomozov@gmail.com>:
> Hi David
>
> On Mon, Aug 7, 2017 at 9:59 AM, David Matlack <dmatlack@google.com> wrote:
>> Likely this instruction is accessing an MMIO region (probably a VGA
>> buffer, given "mov $0xb8f0e,%ecx"). This will cause a trap to KVM
>> and KVM will try to emulate the instruction. The instruction doing the
>> MMIO access is an SSE instruction,
>
> It is correct. The emulation error happens in printf() kernel function
> that formats a string and then prints it VGA area. Compilers (both
> clang and gcc) think that SSE instructions are more efficient in
> dealing with strings and that's why I have SSE instructions here.
>
>> which KVM does not have support to
>> emulate, thus the internal error. If you didn't hand write this
>> assembly I would try compiling your kernel with compilation flags to
>> disable SSE code generation and see if that fixes your issue.
>
> Adding "-mno-sse" and removing some floating point functionality in my
> kernel make the emulation error go away. So it confirms your thesis.
>
> Though I need floating point/SSE in my kernel and I can't disable SSE
> completely. Instead I would love to understand why KVM can't use SSE
> with VGA memory? I tested my kernel binary with bare hardware, KVM,
> VMWare, QEMU and only KVM fails.
Actually the KVM supports SSE emulation in some degree, please post
your testcase here.
Regards,
Wanpeng Li
^ permalink raw reply [flat|nested] 7+ messages in thread
* Re: KVM emulation failure with "movups" instruction
2017-08-07 22:00 ` Wanpeng Li
@ 2017-08-09 22:02 ` Anatol Pomozov
2017-08-09 22:09 ` Wanpeng Li
0 siblings, 1 reply; 7+ messages in thread
From: Anatol Pomozov @ 2017-08-09 22:02 UTC (permalink / raw)
To: Wanpeng Li
Cc: David Matlack, kvm list, Paolo Bonzini, Dan Carpenter, David Hildenbrand
Hello Wanpeng
Unfortunately I do not have a simple test case for this issue. The
issue I see is with a large non-published project compiled with the
latest version of clang.
Isn't information provided by QEMU enough? Is there a way to enable
additional debug log in KVM?
On Mon, Aug 7, 2017 at 3:00 PM, Wanpeng Li <kernellwp@gmail.com> wrote:
> 2017-08-08 2:45 GMT+08:00 Anatol Pomozov <anatol.pomozov@gmail.com>:
>> Hi David
>>
>> On Mon, Aug 7, 2017 at 9:59 AM, David Matlack <dmatlack@google.com> wrote:
>>> Likely this instruction is accessing an MMIO region (probably a VGA
>>> buffer, given "mov $0xb8f0e,%ecx"). This will cause a trap to KVM
>>> and KVM will try to emulate the instruction. The instruction doing the
>>> MMIO access is an SSE instruction,
>>
>> It is correct. The emulation error happens in printf() kernel function
>> that formats a string and then prints it VGA area. Compilers (both
>> clang and gcc) think that SSE instructions are more efficient in
>> dealing with strings and that's why I have SSE instructions here.
>>
>>> which KVM does not have support to
>>> emulate, thus the internal error. If you didn't hand write this
>>> assembly I would try compiling your kernel with compilation flags to
>>> disable SSE code generation and see if that fixes your issue.
>>
>> Adding "-mno-sse" and removing some floating point functionality in my
>> kernel make the emulation error go away. So it confirms your thesis.
>>
>> Though I need floating point/SSE in my kernel and I can't disable SSE
>> completely. Instead I would love to understand why KVM can't use SSE
>> with VGA memory? I tested my kernel binary with bare hardware, KVM,
>> VMWare, QEMU and only KVM fails.
>
> Actually the KVM supports SSE emulation in some degree, please post
> your testcase here.
>
> Regards,
> Wanpeng Li
^ permalink raw reply [flat|nested] 7+ messages in thread
* Re: KVM emulation failure with "movups" instruction
2017-08-09 22:02 ` Anatol Pomozov
@ 2017-08-09 22:09 ` Wanpeng Li
0 siblings, 0 replies; 7+ messages in thread
From: Wanpeng Li @ 2017-08-09 22:09 UTC (permalink / raw)
To: Anatol Pomozov
Cc: David Matlack, kvm list, Paolo Bonzini, Dan Carpenter, David Hildenbrand
2017-08-10 6:02 GMT+08:00 Anatol Pomozov <anatol.pomozov@gmail.com>:
> Hello Wanpeng
>
> Unfortunately I do not have a simple test case for this issue. The
> issue I see is with a large non-published project compiled with the
> latest version of clang.
>
> Isn't information provided by QEMU enough? Is there a way to enable
> additional debug log in KVM?
KVM just has basic support to SSE emulation, and it can't emulate the
movups instruction currently, in addition, I can reproduce this by a
small change in kvm-unit-tests/emulator.flat.
Regards,
Wanpeng Li
>
> On Mon, Aug 7, 2017 at 3:00 PM, Wanpeng Li <kernellwp@gmail.com> wrote:
>> 2017-08-08 2:45 GMT+08:00 Anatol Pomozov <anatol.pomozov@gmail.com>:
>>> Hi David
>>>
>>> On Mon, Aug 7, 2017 at 9:59 AM, David Matlack <dmatlack@google.com> wrote:
>>>> Likely this instruction is accessing an MMIO region (probably a VGA
>>>> buffer, given "mov $0xb8f0e,%ecx"). This will cause a trap to KVM
>>>> and KVM will try to emulate the instruction. The instruction doing the
>>>> MMIO access is an SSE instruction,
>>>
>>> It is correct. The emulation error happens in printf() kernel function
>>> that formats a string and then prints it VGA area. Compilers (both
>>> clang and gcc) think that SSE instructions are more efficient in
>>> dealing with strings and that's why I have SSE instructions here.
>>>
>>>> which KVM does not have support to
>>>> emulate, thus the internal error. If you didn't hand write this
>>>> assembly I would try compiling your kernel with compilation flags to
>>>> disable SSE code generation and see if that fixes your issue.
>>>
>>> Adding "-mno-sse" and removing some floating point functionality in my
>>> kernel make the emulation error go away. So it confirms your thesis.
>>>
>>> Though I need floating point/SSE in my kernel and I can't disable SSE
>>> completely. Instead I would love to understand why KVM can't use SSE
>>> with VGA memory? I tested my kernel binary with bare hardware, KVM,
>>> VMWare, QEMU and only KVM fails.
>>
>> Actually the KVM supports SSE emulation in some degree, please post
>> your testcase here.
>>
>> Regards,
>> Wanpeng Li
^ permalink raw reply [flat|nested] 7+ messages in thread
end of thread, other threads:[~2017-08-09 22:09 UTC | newest]
Thread overview: 7+ messages (download: mbox.gz / follow: Atom feed)
-- links below jump to the message on this page --
2017-08-04 22:03 KVM emulation failure with "movups" instruction Anatol Pomozov
2017-08-04 22:41 ` Wanpeng Li
2017-08-07 16:59 ` David Matlack
2017-08-07 18:45 ` Anatol Pomozov
2017-08-07 22:00 ` Wanpeng Li
2017-08-09 22:02 ` Anatol Pomozov
2017-08-09 22:09 ` Wanpeng Li
This is an external index of several public inboxes,
see mirroring instructions on how to clone and mirror
all data and code used by this external index.