Message ID | 20210402024124.545826-1-npiggin@gmail.com (mailing list archive) |
---|---|
State | Accepted |
Headers | show |
Series | powerpc/powernv: Enable HAIL (HV AIL) for ISA v3.1 processors | expand |
Context | Check | Description |
---|---|---|
snowpatch_ozlabs/apply_patch | success | Successfully applied on branch powerpc/merge (87d76f542a24ecfa797e9bd3bb56c0f19aabff57) |
snowpatch_ozlabs/build-ppc64le | success | Build succeeded |
snowpatch_ozlabs/build-ppc64be | success | Build succeeded |
snowpatch_ozlabs/build-ppc64e | success | Build succeeded |
snowpatch_ozlabs/build-pmac32 | success | Build succeeded |
snowpatch_ozlabs/checkpatch | warning | total: 0 errors, 0 warnings, 1 checks, 33 lines checked |
snowpatch_ozlabs/needsstable | success | Patch has no Fixes tags |
On 4/2/21 4:41 AM, Nicholas Piggin wrote: > Starting with ISA v3.1, LPCR[AIL] no longer controls the interrupt > mode for HV=1 interrupts. Instead, a new LPCR[HAIL] bit is defined > which behaves like AIL=3 for HV interrupts when set. Will QEMU need an update ? Thanks, C. > Set HAIL on bare metal to give us mmu-on interrupts and improve > performance. > > This also fixes an scv bug: we don't implement scv real mode (AIL=0) > vectors because they are at an inconvenient location, so we just > disable scv support when AIL can not be set. However powernv assumes > that LPCR[AIL] will enable AIL mode so it enables scv support despite > HV interrupts being AIL=0, which causes scv interrupts to go off into > the weeds. > > Signed-off-by: Nicholas Piggin <npiggin@gmail.com> > --- > arch/powerpc/include/asm/reg.h | 1 + > arch/powerpc/kernel/setup_64.c | 19 ++++++++++++++++--- > 2 files changed, 17 insertions(+), 3 deletions(-) > > diff --git a/arch/powerpc/include/asm/reg.h b/arch/powerpc/include/asm/reg.h > index 1be20bc8dce2..9086a2644c89 100644 > --- a/arch/powerpc/include/asm/reg.h > +++ b/arch/powerpc/include/asm/reg.h > @@ -441,6 +441,7 @@ > #define LPCR_VRMA_LP1 ASM_CONST(0x0000800000000000) > #define LPCR_RMLS 0x1C000000 /* Implementation dependent RMO limit sel */ > #define LPCR_RMLS_SH 26 > +#define LPCR_HAIL ASM_CONST(0x0000000004000000) /* HV AIL (ISAv3.1) */ > #define LPCR_ILE ASM_CONST(0x0000000002000000) /* !HV irqs set MSR:LE */ > #define LPCR_AIL ASM_CONST(0x0000000001800000) /* Alternate interrupt location */ > #define LPCR_AIL_0 ASM_CONST(0x0000000000000000) /* MMU off exception offset 0x0 */ > diff --git a/arch/powerpc/kernel/setup_64.c b/arch/powerpc/kernel/setup_64.c > index 04a31586f760..671192afcdfd 100644 > --- a/arch/powerpc/kernel/setup_64.c > +++ b/arch/powerpc/kernel/setup_64.c > @@ -233,10 +233,23 @@ static void cpu_ready_for_interrupts(void) > * If we are not in hypervisor mode the job is done once for > * the whole partition in configure_exceptions(). > */ > - if (cpu_has_feature(CPU_FTR_HVMODE) && > - cpu_has_feature(CPU_FTR_ARCH_207S)) { > + if (cpu_has_feature(CPU_FTR_HVMODE)) { > unsigned long lpcr = mfspr(SPRN_LPCR); > - mtspr(SPRN_LPCR, lpcr | LPCR_AIL_3); > + unsigned long new_lpcr = lpcr; > + > + if (cpu_has_feature(CPU_FTR_ARCH_31)) { > + /* P10 DD1 does not have HAIL */ > + if (pvr_version_is(PVR_POWER10) && > + (mfspr(SPRN_PVR) & 0xf00) == 0x100) > + new_lpcr |= LPCR_AIL_3; > + else > + new_lpcr |= LPCR_HAIL; > + } else if (cpu_has_feature(CPU_FTR_ARCH_207S)) { > + new_lpcr |= LPCR_AIL_3; > + } > + > + if (new_lpcr != lpcr) > + mtspr(SPRN_LPCR, new_lpcr); > } > > /* >
Excerpts from Cédric Le Goater's message of April 2, 2021 4:10 pm: > On 4/2/21 4:41 AM, Nicholas Piggin wrote: >> Starting with ISA v3.1, LPCR[AIL] no longer controls the interrupt >> mode for HV=1 interrupts. Instead, a new LPCR[HAIL] bit is defined >> which behaves like AIL=3 for HV interrupts when set. > > Will QEMU need an update ? Yes you're right it will, we need to do that. Thanks, Nick
Nicholas Piggin <npiggin@gmail.com> writes: > Starting with ISA v3.1, LPCR[AIL] no longer controls the interrupt > mode for HV=1 interrupts. Instead, a new LPCR[HAIL] bit is defined > which behaves like AIL=3 for HV interrupts when set. > > Set HAIL on bare metal to give us mmu-on interrupts and improve > performance. > > This also fixes an scv bug: we don't implement scv real mode (AIL=0) > vectors because they are at an inconvenient location, so we just > disable scv support when AIL can not be set. However powernv assumes > that LPCR[AIL] will enable AIL mode so it enables scv support despite > HV interrupts being AIL=0, which causes scv interrupts to go off into > the weeds. Should we tag this as fixing the initial P10 support, or the scv support? Or neither? cheers > diff --git a/arch/powerpc/include/asm/reg.h b/arch/powerpc/include/asm/reg.h > index 1be20bc8dce2..9086a2644c89 100644 > --- a/arch/powerpc/include/asm/reg.h > +++ b/arch/powerpc/include/asm/reg.h > @@ -441,6 +441,7 @@ > #define LPCR_VRMA_LP1 ASM_CONST(0x0000800000000000) > #define LPCR_RMLS 0x1C000000 /* Implementation dependent RMO limit sel */ > #define LPCR_RMLS_SH 26 > +#define LPCR_HAIL ASM_CONST(0x0000000004000000) /* HV AIL (ISAv3.1) */ > #define LPCR_ILE ASM_CONST(0x0000000002000000) /* !HV irqs set MSR:LE */ > #define LPCR_AIL ASM_CONST(0x0000000001800000) /* Alternate interrupt location */ > #define LPCR_AIL_0 ASM_CONST(0x0000000000000000) /* MMU off exception offset 0x0 */ > diff --git a/arch/powerpc/kernel/setup_64.c b/arch/powerpc/kernel/setup_64.c > index 04a31586f760..671192afcdfd 100644 > --- a/arch/powerpc/kernel/setup_64.c > +++ b/arch/powerpc/kernel/setup_64.c > @@ -233,10 +233,23 @@ static void cpu_ready_for_interrupts(void) > * If we are not in hypervisor mode the job is done once for > * the whole partition in configure_exceptions(). > */ > - if (cpu_has_feature(CPU_FTR_HVMODE) && > - cpu_has_feature(CPU_FTR_ARCH_207S)) { > + if (cpu_has_feature(CPU_FTR_HVMODE)) { > unsigned long lpcr = mfspr(SPRN_LPCR); > - mtspr(SPRN_LPCR, lpcr | LPCR_AIL_3); > + unsigned long new_lpcr = lpcr; > + > + if (cpu_has_feature(CPU_FTR_ARCH_31)) { > + /* P10 DD1 does not have HAIL */ > + if (pvr_version_is(PVR_POWER10) && > + (mfspr(SPRN_PVR) & 0xf00) == 0x100) > + new_lpcr |= LPCR_AIL_3; > + else > + new_lpcr |= LPCR_HAIL; > + } else if (cpu_has_feature(CPU_FTR_ARCH_207S)) { > + new_lpcr |= LPCR_AIL_3; > + } > + > + if (new_lpcr != lpcr) > + mtspr(SPRN_LPCR, new_lpcr); > } > > /* > -- > 2.23.0
Excerpts from Michael Ellerman's message of April 7, 2021 9:33 pm: > Nicholas Piggin <npiggin@gmail.com> writes: >> Starting with ISA v3.1, LPCR[AIL] no longer controls the interrupt >> mode for HV=1 interrupts. Instead, a new LPCR[HAIL] bit is defined >> which behaves like AIL=3 for HV interrupts when set. >> >> Set HAIL on bare metal to give us mmu-on interrupts and improve >> performance. >> >> This also fixes an scv bug: we don't implement scv real mode (AIL=0) >> vectors because they are at an inconvenient location, so we just >> disable scv support when AIL can not be set. However powernv assumes >> that LPCR[AIL] will enable AIL mode so it enables scv support despite >> HV interrupts being AIL=0, which causes scv interrupts to go off into >> the weeds. > > Should we tag this as fixing the initial P10 support, or the scv > support? Or neither? Good question. It does fix a nasty crash with scv so at least it should be tagged I guess. I don't know of anything else that assumes AIL on bare metal, so I don't know of a crashy bug it fixes with initial P10 support. But it is a bit odd for a HV OS running a v3.1 processor to set the old LPCR AIL bits, so it is some kind of bug fix (performance at least). Could go either way I guess. Thanks, Nick > > cheers > >> diff --git a/arch/powerpc/include/asm/reg.h b/arch/powerpc/include/asm/reg.h >> index 1be20bc8dce2..9086a2644c89 100644 >> --- a/arch/powerpc/include/asm/reg.h >> +++ b/arch/powerpc/include/asm/reg.h >> @@ -441,6 +441,7 @@ >> #define LPCR_VRMA_LP1 ASM_CONST(0x0000800000000000) >> #define LPCR_RMLS 0x1C000000 /* Implementation dependent RMO limit sel */ >> #define LPCR_RMLS_SH 26 >> +#define LPCR_HAIL ASM_CONST(0x0000000004000000) /* HV AIL (ISAv3.1) */ >> #define LPCR_ILE ASM_CONST(0x0000000002000000) /* !HV irqs set MSR:LE */ >> #define LPCR_AIL ASM_CONST(0x0000000001800000) /* Alternate interrupt location */ >> #define LPCR_AIL_0 ASM_CONST(0x0000000000000000) /* MMU off exception offset 0x0 */ >> diff --git a/arch/powerpc/kernel/setup_64.c b/arch/powerpc/kernel/setup_64.c >> index 04a31586f760..671192afcdfd 100644 >> --- a/arch/powerpc/kernel/setup_64.c >> +++ b/arch/powerpc/kernel/setup_64.c >> @@ -233,10 +233,23 @@ static void cpu_ready_for_interrupts(void) >> * If we are not in hypervisor mode the job is done once for >> * the whole partition in configure_exceptions(). >> */ >> - if (cpu_has_feature(CPU_FTR_HVMODE) && >> - cpu_has_feature(CPU_FTR_ARCH_207S)) { >> + if (cpu_has_feature(CPU_FTR_HVMODE)) { >> unsigned long lpcr = mfspr(SPRN_LPCR); >> - mtspr(SPRN_LPCR, lpcr | LPCR_AIL_3); >> + unsigned long new_lpcr = lpcr; >> + >> + if (cpu_has_feature(CPU_FTR_ARCH_31)) { >> + /* P10 DD1 does not have HAIL */ >> + if (pvr_version_is(PVR_POWER10) && >> + (mfspr(SPRN_PVR) & 0xf00) == 0x100) >> + new_lpcr |= LPCR_AIL_3; >> + else >> + new_lpcr |= LPCR_HAIL; >> + } else if (cpu_has_feature(CPU_FTR_ARCH_207S)) { >> + new_lpcr |= LPCR_AIL_3; >> + } >> + >> + if (new_lpcr != lpcr) >> + mtspr(SPRN_LPCR, new_lpcr); >> } >> >> /* >> -- >> 2.23.0 >
On Fri, 2 Apr 2021 12:41:24 +1000, Nicholas Piggin wrote: > Starting with ISA v3.1, LPCR[AIL] no longer controls the interrupt > mode for HV=1 interrupts. Instead, a new LPCR[HAIL] bit is defined > which behaves like AIL=3 for HV interrupts when set. > > Set HAIL on bare metal to give us mmu-on interrupts and improve > performance. > > [...] Applied to powerpc/next. [1/1] powerpc/powernv: Enable HAIL (HV AIL) for ISA v3.1 processors https://git.kernel.org/powerpc/c/49c1d07fd04f54eb588c4a1dfcedc8d22c5ffd50 cheers
diff --git a/arch/powerpc/include/asm/reg.h b/arch/powerpc/include/asm/reg.h index 1be20bc8dce2..9086a2644c89 100644 --- a/arch/powerpc/include/asm/reg.h +++ b/arch/powerpc/include/asm/reg.h @@ -441,6 +441,7 @@ #define LPCR_VRMA_LP1 ASM_CONST(0x0000800000000000) #define LPCR_RMLS 0x1C000000 /* Implementation dependent RMO limit sel */ #define LPCR_RMLS_SH 26 +#define LPCR_HAIL ASM_CONST(0x0000000004000000) /* HV AIL (ISAv3.1) */ #define LPCR_ILE ASM_CONST(0x0000000002000000) /* !HV irqs set MSR:LE */ #define LPCR_AIL ASM_CONST(0x0000000001800000) /* Alternate interrupt location */ #define LPCR_AIL_0 ASM_CONST(0x0000000000000000) /* MMU off exception offset 0x0 */ diff --git a/arch/powerpc/kernel/setup_64.c b/arch/powerpc/kernel/setup_64.c index 04a31586f760..671192afcdfd 100644 --- a/arch/powerpc/kernel/setup_64.c +++ b/arch/powerpc/kernel/setup_64.c @@ -233,10 +233,23 @@ static void cpu_ready_for_interrupts(void) * If we are not in hypervisor mode the job is done once for * the whole partition in configure_exceptions(). */ - if (cpu_has_feature(CPU_FTR_HVMODE) && - cpu_has_feature(CPU_FTR_ARCH_207S)) { + if (cpu_has_feature(CPU_FTR_HVMODE)) { unsigned long lpcr = mfspr(SPRN_LPCR); - mtspr(SPRN_LPCR, lpcr | LPCR_AIL_3); + unsigned long new_lpcr = lpcr; + + if (cpu_has_feature(CPU_FTR_ARCH_31)) { + /* P10 DD1 does not have HAIL */ + if (pvr_version_is(PVR_POWER10) && + (mfspr(SPRN_PVR) & 0xf00) == 0x100) + new_lpcr |= LPCR_AIL_3; + else + new_lpcr |= LPCR_HAIL; + } else if (cpu_has_feature(CPU_FTR_ARCH_207S)) { + new_lpcr |= LPCR_AIL_3; + } + + if (new_lpcr != lpcr) + mtspr(SPRN_LPCR, new_lpcr); } /*
Starting with ISA v3.1, LPCR[AIL] no longer controls the interrupt mode for HV=1 interrupts. Instead, a new LPCR[HAIL] bit is defined which behaves like AIL=3 for HV interrupts when set. Set HAIL on bare metal to give us mmu-on interrupts and improve performance. This also fixes an scv bug: we don't implement scv real mode (AIL=0) vectors because they are at an inconvenient location, so we just disable scv support when AIL can not be set. However powernv assumes that LPCR[AIL] will enable AIL mode so it enables scv support despite HV interrupts being AIL=0, which causes scv interrupts to go off into the weeds. Signed-off-by: Nicholas Piggin <npiggin@gmail.com> --- arch/powerpc/include/asm/reg.h | 1 + arch/powerpc/kernel/setup_64.c | 19 ++++++++++++++++--- 2 files changed, 17 insertions(+), 3 deletions(-)