* Re: [kvm-ppc-devel] [PATCH 1/5] Remove KVM_HANDLER macro.
2008-01-25 7:12 [kvm-ppc-devel] [PATCH 1/5] Remove KVM_HANDLER macro Zhang Wei
@ 2008-01-28 20:41 ` Hollis Blanchard
2008-01-29 4:16 ` Hollis Blanchard
` (3 subsequent siblings)
4 siblings, 0 replies; 6+ messages in thread
From: Hollis Blanchard @ 2008-01-28 20:41 UTC (permalink / raw)
To: kvm-ppc
On Fri, 2008-01-25 at 15:12 +0800, Zhang Wei wrote:
> Replace KVM_HANDLER macro by kvm_trampoline_handler() function.
> Codes define by KVM_HANDLER marco will be extracted by
> kvm_trampoline_handler install loop.
I got rid of the trampoline stuff in the last patch I sent out, so this
doesn't apply any more.
--
Hollis Blanchard
IBM Linux Technology Center
-------------------------------------------------------------------------
This SF.net email is sponsored by: Microsoft
Defy all challenges. Microsoft(R) Visual Studio 2008.
http://clk.atdmt.com/MRT/go/vse0120000070mrt/direct/01/
_______________________________________________
kvm-ppc-devel mailing list
kvm-ppc-devel@lists.sourceforge.net
https://lists.sourceforge.net/lists/listinfo/kvm-ppc-devel
^ permalink raw reply [flat|nested] 6+ messages in thread* Re: [kvm-ppc-devel] [PATCH 1/5] Remove KVM_HANDLER macro.
2008-01-25 7:12 [kvm-ppc-devel] [PATCH 1/5] Remove KVM_HANDLER macro Zhang Wei
2008-01-28 20:41 ` Hollis Blanchard
@ 2008-01-29 4:16 ` Hollis Blanchard
2008-01-29 6:02 ` Zhang Wei
` (2 subsequent siblings)
4 siblings, 0 replies; 6+ messages in thread
From: Hollis Blanchard @ 2008-01-29 4:16 UTC (permalink / raw)
To: kvm-ppc
On Fri, 2008-01-25 at 15:12 +0800, Zhang Wei wrote:
> Replace KVM_HANDLER macro by kvm_trampoline_handler() function.
> Codes define by KVM_HANDLER marco will be extracted by
> kvm_trampoline_handler install loop.
>
> Signed-off-by: Zhang Wei <wei.zhang@freescale.com>
> ---
> arch/powerpc/kvm/booke_interrupts.S | 32 ++++++--------------------------
> arch/powerpc/kvm/powerpc.c | 6 +++++-
> include/asm-powerpc/kvm_host.h | 1 +
> 3 files changed, 12 insertions(+), 27 deletions(-)
>
> diff --git a/arch/powerpc/kvm/booke_interrupts.S b/arch/powerpc/kvm/booke_interrupts.S
> index 45131cc..ec5f071 100644
> --- a/arch/powerpc/kvm/booke_interrupts.S
> +++ b/arch/powerpc/kvm/booke_interrupts.S
> @@ -39,36 +39,19 @@
> #define HOST_STACK_SIZE (((HOST_MIN_STACK_SIZE + 15) / 16) * 16) /* Align. */
> #define HOST_STACK_LR (HOST_STACK_SIZE + 4) /* In caller stack frame. */
>
> -.macro KVM_HANDLER ivor_nr
> -_GLOBAL(kvm_trampoline_handler_\ivor_nr)
> +_GLOBAL(kvm_trampoline_handler)
> /* Get pointer to vcpu and record exit number. */
> mtspr SPRN_SPRG0, r4
> mfspr r4, SPRN_SPRG1
> stw r5, VCPU_GPR(r5)(r4)
> - li r5, \ivor_nr
> + li r5, 0 /* The SIMM field of this instruction will be replaced
> + * at install time by ivor_nr.
> + */
> /* This branch is fixed up at install time to jump to
> * kvm_trampoline_resume_host(). */
> b .
> -.endm
> -
> -_GLOBAL(kvm_trampoline_start)
> -
> -KVM_HANDLER PPC44x_INTERRUPT_CRITICAL
> -KVM_HANDLER PPC44x_INTERRUPT_MACHINE_CHECK
> -KVM_HANDLER PPC44x_INTERRUPT_DATA_STORAGE
> -KVM_HANDLER PPC44x_INTERRUPT_INST_STORAGE
> -KVM_HANDLER PPC44x_INTERRUPT_EXTERNAL
> -KVM_HANDLER PPC44x_INTERRUPT_ALIGNMENT
> -KVM_HANDLER PPC44x_INTERRUPT_PROGRAM
> -KVM_HANDLER PPC44x_INTERRUPT_FP_UNAVAIL
> -KVM_HANDLER PPC44x_INTERRUPT_SYSCALL
> -KVM_HANDLER PPC44x_INTERRUPT_AP_UNAVAIL
> -KVM_HANDLER PPC44x_INTERRUPT_DECREMENTER
> -KVM_HANDLER PPC44x_INTERRUPT_FIT
> -KVM_HANDLER PPC44x_INTERRUPT_WATCHDOG
> -KVM_HANDLER PPC44x_INTERRUPT_DTLB_MISS
> -KVM_HANDLER PPC44x_INTERRUPT_ITLB_MISS
> -KVM_HANDLER PPC44x_INTERRUPT_DEBUG
> +_GLOBAL(kvm_trampoline_handler_len)
> + .long . - kvm_trampoline_handler
>
> #define NEED_INST_MASK ((1<<PPC44x_INTERRUPT_PROGRAM) | \
> (1<<PPC44x_INTERRUPT_DTLB_MISS))
> @@ -208,9 +191,6 @@ _GLOBAL(kvm_trampoline_resume_guest)
> _GLOBAL(kvm_trampoline_resume_guest_len)
> .long . - kvm_trampoline_resume_guest
>
> -_GLOBAL(kvm_trampoline_handler_len)
> - .long kvm_trampoline_handler_1 - kvm_trampoline_handler_0
> -
>
> /* Registers:
> * SPRG0: guest r4
> diff --git a/arch/powerpc/kvm/powerpc.c b/arch/powerpc/kvm/powerpc.c
> index cbdacd5..0839cfc 100644
> --- a/arch/powerpc/kvm/powerpc.c
> +++ b/arch/powerpc/kvm/powerpc.c
> @@ -466,7 +466,7 @@ struct kvm_vcpu *kvm_arch_vcpu_create(struct kvm *kvm, unsigned int id)
> ivorlist[15] = mfspr(SPRN_IVOR15);
> for (i = 0; i < 16; i++) {
> memcpy(handlers + ivorlist[i],
> - kvm_trampoline_start + i * kvm_trampoline_handler_len,
> + &kvm_trampoline_handler,
> kvm_trampoline_handler_len);
> }
>
> @@ -484,9 +484,13 @@ struct kvm_vcpu *kvm_arch_vcpu_create(struct kvm *kvm, unsigned int id)
> /* Manually fix up the handler branches, since we moved the code away
> * from its link address. */
> for (i = 0; i < 16; i++) {
> + unsigned long *lir5;
> unsigned long *branch;
> + lir5 = handlers + ivorlist[i] + kvm_trampoline_handler_len
> + - 8;
> branch = handlers + ivorlist[i] + kvm_trampoline_handler_len
> - 4;
> + *lir5 |= i & 0xffff;
> *branch |= resume_host - (void *)branch;
> }
>
> diff --git a/include/asm-powerpc/kvm_host.h b/include/asm-powerpc/kvm_host.h
> index 088780f..ef57aec 100644
> --- a/include/asm-powerpc/kvm_host.h
> +++ b/include/asm-powerpc/kvm_host.h
> @@ -143,6 +143,7 @@ enum emulation_result {
>
> extern int __kvmppc_vcpu_run(struct kvm_run *kvm_run, struct kvm_vcpu *vcpu);
>
> +extern unsigned long *kvm_trampoline_handler;
> extern char kvm_trampoline_start[];
> extern void kvm_trampoline_resume_host(void);
> extern unsigned long kvm_trampoline_resume_host_len;
Actually, this patch could still apply with some renaming.
However, now that we've gotten away from binary patching the branch
instruction, I'm not eager to go back to that with the "li" instruction.
Each handler is only 24 bytes, so I'm not that worried about duplicating
them, and the readability and simplicity are better with the
duplication.
--
Hollis Blanchard
IBM Linux Technology Center
-------------------------------------------------------------------------
This SF.net email is sponsored by: Microsoft
Defy all challenges. Microsoft(R) Visual Studio 2008.
http://clk.atdmt.com/MRT/go/vse0120000070mrt/direct/01/
_______________________________________________
kvm-ppc-devel mailing list
kvm-ppc-devel@lists.sourceforge.net
https://lists.sourceforge.net/lists/listinfo/kvm-ppc-devel
^ permalink raw reply [flat|nested] 6+ messages in thread* Re: [kvm-ppc-devel] [PATCH 1/5] Remove KVM_HANDLER macro.
2008-01-25 7:12 [kvm-ppc-devel] [PATCH 1/5] Remove KVM_HANDLER macro Zhang Wei
2008-01-28 20:41 ` Hollis Blanchard
2008-01-29 4:16 ` Hollis Blanchard
@ 2008-01-29 6:02 ` Zhang Wei
2008-01-29 16:41 ` Hollis Blanchard
2008-01-30 5:46 ` Zhang Wei
4 siblings, 0 replies; 6+ messages in thread
From: Zhang Wei @ 2008-01-29 6:02 UTC (permalink / raw)
To: kvm-ppc
Hi, Hollis,
> -----Original Message-----
> From: Hollis Blanchard [mailto:hollisb@us.ibm.com]
> Sent: Tuesday, January 29, 2008 12:16 PM
> To: Zhang Wei
> Cc: kvm-ppc-devel@lists.sourceforge.net
> Subject: Re: [kvm-ppc-devel] [PATCH 1/5] Remove KVM_HANDLER macro.
>
> On Fri, 2008-01-25 at 15:12 +0800, Zhang Wei wrote:
> > Replace KVM_HANDLER macro by kvm_trampoline_handler() function.
> > Codes define by KVM_HANDLER marco will be extracted by
> > kvm_trampoline_handler install loop.
> >
> > Signed-off-by: Zhang Wei <wei.zhang@freescale.com>
> > ---
> > arch/powerpc/kvm/booke_interrupts.S | 32
> ++++++--------------------------
> > arch/powerpc/kvm/powerpc.c | 6 +++++-
> > include/asm-powerpc/kvm_host.h | 1 +
> > 3 files changed, 12 insertions(+), 27 deletions(-)
> >
> > diff --git a/arch/powerpc/kvm/booke_interrupts.S
> b/arch/powerpc/kvm/booke_interrupts.S
> > index 45131cc..ec5f071 100644
> > --- a/arch/powerpc/kvm/booke_interrupts.S
> > +++ b/arch/powerpc/kvm/booke_interrupts.S
> > @@ -39,36 +39,19 @@
> > #define HOST_STACK_SIZE (((HOST_MIN_STACK_SIZE + 15) / 16)
> * 16) /* Align. */
> > #define HOST_STACK_LR (HOST_STACK_SIZE + 4) /* In caller
> stack frame. */
> >
> > -.macro KVM_HANDLER ivor_nr
> > -_GLOBAL(kvm_trampoline_handler_\ivor_nr)
> > +_GLOBAL(kvm_trampoline_handler)
> > /* Get pointer to vcpu and record exit number. */
> > mtspr SPRN_SPRG0, r4
> > mfspr r4, SPRN_SPRG1
> > stw r5, VCPU_GPR(r5)(r4)
> > - li r5, \ivor_nr
> > + li r5, 0 /* The SIMM field of this instruction
> will be replaced
> > + * at install time by ivor_nr.
> > + */
> > /* This branch is fixed up at install time to jump to
> > * kvm_trampoline_resume_host(). */
> > b .
> > -.endm
> > -
> > -_GLOBAL(kvm_trampoline_start)
> > -
> > -KVM_HANDLER PPC44x_INTERRUPT_CRITICAL
> > -KVM_HANDLER PPC44x_INTERRUPT_MACHINE_CHECK
> > -KVM_HANDLER PPC44x_INTERRUPT_DATA_STORAGE
> > -KVM_HANDLER PPC44x_INTERRUPT_INST_STORAGE
> > -KVM_HANDLER PPC44x_INTERRUPT_EXTERNAL
> > -KVM_HANDLER PPC44x_INTERRUPT_ALIGNMENT
> > -KVM_HANDLER PPC44x_INTERRUPT_PROGRAM
> > -KVM_HANDLER PPC44x_INTERRUPT_FP_UNAVAIL
> > -KVM_HANDLER PPC44x_INTERRUPT_SYSCALL
> > -KVM_HANDLER PPC44x_INTERRUPT_AP_UNAVAIL
> > -KVM_HANDLER PPC44x_INTERRUPT_DECREMENTER
> > -KVM_HANDLER PPC44x_INTERRUPT_FIT
> > -KVM_HANDLER PPC44x_INTERRUPT_WATCHDOG
> > -KVM_HANDLER PPC44x_INTERRUPT_DTLB_MISS
> > -KVM_HANDLER PPC44x_INTERRUPT_ITLB_MISS
> > -KVM_HANDLER PPC44x_INTERRUPT_DEBUG
> > +_GLOBAL(kvm_trampoline_handler_len)
> > + .long . - kvm_trampoline_handler
> >
> > #define NEED_INST_MASK ((1<<PPC44x_INTERRUPT_PROGRAM) | \
> > (1<<PPC44x_INTERRUPT_DTLB_MISS))
> > @@ -208,9 +191,6 @@ _GLOBAL(kvm_trampoline_resume_guest)
> > _GLOBAL(kvm_trampoline_resume_guest_len)
> > .long . - kvm_trampoline_resume_guest
> >
> > -_GLOBAL(kvm_trampoline_handler_len)
> > - .long kvm_trampoline_handler_1 - kvm_trampoline_handler_0
> > -
> >
> > /* Registers:
> > * SPRG0: guest r4
> > diff --git a/arch/powerpc/kvm/powerpc.c b/arch/powerpc/kvm/powerpc.c
> > index cbdacd5..0839cfc 100644
> > --- a/arch/powerpc/kvm/powerpc.c
> > +++ b/arch/powerpc/kvm/powerpc.c
> > @@ -466,7 +466,7 @@ struct kvm_vcpu
> *kvm_arch_vcpu_create(struct kvm *kvm, unsigned int id)
> > ivorlist[15] = mfspr(SPRN_IVOR15);
> > for (i = 0; i < 16; i++) {
> > memcpy(handlers + ivorlist[i],
> > - kvm_trampoline_start + i *
> kvm_trampoline_handler_len,
> > + &kvm_trampoline_handler,
> > kvm_trampoline_handler_len);
> > }
> >
> > @@ -484,9 +484,13 @@ struct kvm_vcpu
> *kvm_arch_vcpu_create(struct kvm *kvm, unsigned int id)
> > /* Manually fix up the handler branches, since we moved
> the code away
> > * from its link address. */
> > for (i = 0; i < 16; i++) {
> > + unsigned long *lir5;
> > unsigned long *branch;
> > + lir5 = handlers + ivorlist[i] +
> kvm_trampoline_handler_len
> > + - 8;
> > branch = handlers + ivorlist[i] +
> kvm_trampoline_handler_len
> > - 4;
> > + *lir5 |= i & 0xffff;
> > *branch |= resume_host - (void *)branch;
> > }
> >
> > diff --git a/include/asm-powerpc/kvm_host.h
> b/include/asm-powerpc/kvm_host.h
> > index 088780f..ef57aec 100644
> > --- a/include/asm-powerpc/kvm_host.h
> > +++ b/include/asm-powerpc/kvm_host.h
> > @@ -143,6 +143,7 @@ enum emulation_result {
> >
> > extern int __kvmppc_vcpu_run(struct kvm_run *kvm_run,
> struct kvm_vcpu *vcpu);
> >
> > +extern unsigned long *kvm_trampoline_handler;
> > extern char kvm_trampoline_start[];
> > extern void kvm_trampoline_resume_host(void);
> > extern unsigned long kvm_trampoline_resume_host_len;
>
> Actually, this patch could still apply with some renaming.
>
> However, now that we've gotten away from binary patching the branch
> instruction, I'm not eager to go back to that with the "li"
> instruction.
> Each handler is only 24 bytes, so I'm not that worried about
> duplicating
> them, and the readability and simplicity are better with the
> duplication.
>
Another reason is e500 have additional exceptions for SPE exceptions
(IVOR32 - IVOR35).
And there are hole from IVOR16 to IVOR 31. If we use fixed macro
functions definition,
we will leave blank functions there. How about it?
Cheers!
Wei
-------------------------------------------------------------------------
This SF.net email is sponsored by: Microsoft
Defy all challenges. Microsoft(R) Visual Studio 2008.
http://clk.atdmt.com/MRT/go/vse0120000070mrt/direct/01/
_______________________________________________
kvm-ppc-devel mailing list
kvm-ppc-devel@lists.sourceforge.net
https://lists.sourceforge.net/lists/listinfo/kvm-ppc-devel
^ permalink raw reply [flat|nested] 6+ messages in thread* Re: [kvm-ppc-devel] [PATCH 1/5] Remove KVM_HANDLER macro.
2008-01-25 7:12 [kvm-ppc-devel] [PATCH 1/5] Remove KVM_HANDLER macro Zhang Wei
` (2 preceding siblings ...)
2008-01-29 6:02 ` Zhang Wei
@ 2008-01-29 16:41 ` Hollis Blanchard
2008-01-30 5:46 ` Zhang Wei
4 siblings, 0 replies; 6+ messages in thread
From: Hollis Blanchard @ 2008-01-29 16:41 UTC (permalink / raw)
To: kvm-ppc
On Tue, 2008-01-29 at 14:02 +0800, Zhang Wei wrote:
>
> > However, now that we've gotten away from binary patching the branch
> > instruction, I'm not eager to go back to that with the "li"
> > instruction.
> > Each handler is only 24 bytes, so I'm not that worried about
> > duplicating
> > them, and the readability and simplicity are better with the
> > duplication.
> >
>
> Another reason is e500 have additional exceptions for SPE exceptions
> (IVOR32 - IVOR35).
> And there are hole from IVOR16 to IVOR 31. If we use fixed macro
> functions definition,
> we will leave blank functions there. How about it?
True, in that case I think your approach will be best. Could you rebase
the patch?
--
Hollis Blanchard
IBM Linux Technology Center
-------------------------------------------------------------------------
This SF.net email is sponsored by: Microsoft
Defy all challenges. Microsoft(R) Visual Studio 2008.
http://clk.atdmt.com/MRT/go/vse0120000070mrt/direct/01/
_______________________________________________
kvm-ppc-devel mailing list
kvm-ppc-devel@lists.sourceforge.net
https://lists.sourceforge.net/lists/listinfo/kvm-ppc-devel
^ permalink raw reply [flat|nested] 6+ messages in thread* Re: [kvm-ppc-devel] [PATCH 1/5] Remove KVM_HANDLER macro.
2008-01-25 7:12 [kvm-ppc-devel] [PATCH 1/5] Remove KVM_HANDLER macro Zhang Wei
` (3 preceding siblings ...)
2008-01-29 16:41 ` Hollis Blanchard
@ 2008-01-30 5:46 ` Zhang Wei
4 siblings, 0 replies; 6+ messages in thread
From: Zhang Wei @ 2008-01-30 5:46 UTC (permalink / raw)
To: kvm-ppc
Hi, Hollis,
I'll rebase the patch. And I've put your new patch(01/24) into the tree.
:)
Cheers!
Wei.
> -----Original Message-----
> From: Hollis Blanchard [mailto:hollisb@us.ibm.com]
> Sent: Wednesday, January 30, 2008 12:42 AM
> To: Zhang Wei
> Cc: kvm-ppc-devel@lists.sourceforge.net
> Subject: RE: [kvm-ppc-devel] [PATCH 1/5] Remove KVM_HANDLER macro.
>
> On Tue, 2008-01-29 at 14:02 +0800, Zhang Wei wrote:
> >
> > > However, now that we've gotten away from binary patching
> the branch
> > > instruction, I'm not eager to go back to that with the "li"
> > > instruction.
> > > Each handler is only 24 bytes, so I'm not that worried about
> > > duplicating
> > > them, and the readability and simplicity are better with the
> > > duplication.
> > >
> >
> > Another reason is e500 have additional exceptions for SPE exceptions
> > (IVOR32 - IVOR35).
> > And there are hole from IVOR16 to IVOR 31. If we use fixed macro
> > functions definition,
> > we will leave blank functions there. How about it?
>
> True, in that case I think your approach will be best. Could
> you rebase
> the patch?
>
> --
> Hollis Blanchard
> IBM Linux Technology Center
>
>
-------------------------------------------------------------------------
This SF.net email is sponsored by: Microsoft
Defy all challenges. Microsoft(R) Visual Studio 2008.
http://clk.atdmt.com/MRT/go/vse0120000070mrt/direct/01/
_______________________________________________
kvm-ppc-devel mailing list
kvm-ppc-devel@lists.sourceforge.net
https://lists.sourceforge.net/lists/listinfo/kvm-ppc-devel
^ permalink raw reply [flat|nested] 6+ messages in thread