| Message ID | 20170216050339.GC19110@fergus.ozlabs.ibm.com (mailing list archive) |
|---|---|
| State | Accepted |
| Commit | dbfcf3cb9c681aa0c5d0bb46068f98d5b1823dd3 |
| Headers | show |
On Thu, 2017-02-16 at 16:03 +1100, Paul Mackerras wrote: > On POWER9, since commit cc3d2940133d ("powerpc/64: Enable use of > radix > MMU under hypervisor on POWER9", 2017-01-30), we set both the radix > and > HPT bits in the client-architecture-support (CAS) vector, which tells > the hypervisor that we can do either radix or HPT. According to > PAPR, > if we use this combination we are promising to do a > H_REGISTER_PROC_TBL > hcall later on to let the hypervisor know whether we are doing radix > or HPT. We currently do this call if we are doing radix but not if > we are doing HPT. If the hypervisor is able to support both radix > and HPT guests, it would be entitled to defer allocation of the HPT > until the H_REGISTER_PROC_TBL call, and to fail any attempts to > create > HPTEs until the H_REGISTER_PROC_TBL call. Thus we need to do a > H_REGISTER_PROC_TBL call when we are doing HPT; otherwise we may > crash at boot time. > > This adds the code to call H_REGISTER_PROC_TBL in this case, before > we attempt to create any HPT entries using H_ENTER. > > Fixes: cc3d2940133d ("powerpc/64: Enable use of radix MMU under > hypervisor on POWER9") > Signed-off-by: Paul Mackerras <paulus@ozlabs.org> > --- > This needs to go in after the topic/ppc-kvm branch. > > arch/powerpc/mm/hash_utils_64.c | 6 ++++++ > arch/powerpc/platforms/pseries/lpar.c | 8 ++++++-- > 2 files changed, 12 insertions(+), 2 deletions(-) > > diff --git a/arch/powerpc/mm/hash_utils_64.c > b/arch/powerpc/mm/hash_utils_64.c > index 8033493..b0ed96e 100644 > --- a/arch/powerpc/mm/hash_utils_64.c > +++ b/arch/powerpc/mm/hash_utils_64.c > @@ -839,6 +839,12 @@ static void __init htab_initialize(void) > /* Using a hypervisor which owns the htab */ > htab_address = NULL; > _SDR1 = 0; > + /* > + * On POWER9, we need to do a H_REGISTER_PROC_TBL > hcall > + * to inform the hypervisor that we wish to use the > HPT. > + */ > + if (cpu_has_feature(CPU_FTR_ARCH_300)) > + register_process_table(0, 0, 0); > #ifdef CONFIG_FA_DUMP > /* > * If firmware assisted dump is active firmware > preserves > diff --git a/arch/powerpc/platforms/pseries/lpar.c > b/arch/powerpc/platforms/pseries/lpar.c > index 0587655..5b47026 100644 > --- a/arch/powerpc/platforms/pseries/lpar.c > +++ b/arch/powerpc/platforms/pseries/lpar.c > @@ -609,15 +609,18 @@ static int __init disable_bulk_remove(char > *str) > > __setup("bulk_remove=", disable_bulk_remove); > > -/* Actually only used for radix, so far */ > static int pseries_lpar_register_process_table(unsigned long base, > unsigned long page_size, unsigned long > table_size) > { > long rc; > - unsigned long flags = PROC_TABLE_NEW; > + unsigned long flags = 0; > > + if (table_size) > + flags |= PROC_TABLE_NEW; > if (radix_enabled()) > flags |= PROC_TABLE_RADIX | PROC_TABLE_GTSE; > + else > + flags |= PROC_TABLE_HPT_SLB; > for (;;) { > rc = plpar_hcall_norets(H_REGISTER_PROC_TBL, flags, > base, > page_size, table_size); > @@ -643,6 +646,7 @@ void __init hpte_init_pseries(void) > mmu_hash_ops.flush_hash_range = > pSeries_lpar_flush_hash_range; > mmu_hash_ops.hpte_clear_all = pseries_hpte_clear_all; > mmu_hash_ops.hugepage_invalidate = > pSeries_lpar_hugepage_invalidate; > + register_process_table = > pseries_lpar_register_process_table; > } > > void radix_init_pseries(void) FWIW: Reviewed-by: Suraj Jitindar Singh <sjitindarsingh@gmail.com>
On Thu, 2017-02-16 at 05:03:39 UTC, Paul Mackerras wrote: > On POWER9, since commit cc3d2940133d ("powerpc/64: Enable use of radix > MMU under hypervisor on POWER9", 2017-01-30), we set both the radix and > HPT bits in the client-architecture-support (CAS) vector, which tells > the hypervisor that we can do either radix or HPT. According to PAPR, > if we use this combination we are promising to do a H_REGISTER_PROC_TBL > hcall later on to let the hypervisor know whether we are doing radix > or HPT. We currently do this call if we are doing radix but not if > we are doing HPT. If the hypervisor is able to support both radix > and HPT guests, it would be entitled to defer allocation of the HPT > until the H_REGISTER_PROC_TBL call, and to fail any attempts to create > HPTEs until the H_REGISTER_PROC_TBL call. Thus we need to do a > H_REGISTER_PROC_TBL call when we are doing HPT; otherwise we may > crash at boot time. > > This adds the code to call H_REGISTER_PROC_TBL in this case, before > we attempt to create any HPT entries using H_ENTER. > > Fixes: cc3d2940133d ("powerpc/64: Enable use of radix MMU under hypervisor on POWER9") > Signed-off-by: Paul Mackerras <paulus@ozlabs.org> > Reviewed-by: Suraj Jitindar Singh <sjitindarsingh@gmail.com> Applied to powerpc next, thanks. https://git.kernel.org/powerpc/c/dbfcf3cb9c681aa0c5d0bb46068f98 cheers
diff --git a/arch/powerpc/mm/hash_utils_64.c b/arch/powerpc/mm/hash_utils_64.c index 8033493..b0ed96e 100644 --- a/arch/powerpc/mm/hash_utils_64.c +++ b/arch/powerpc/mm/hash_utils_64.c @@ -839,6 +839,12 @@ static void __init htab_initialize(void) /* Using a hypervisor which owns the htab */ htab_address = NULL; _SDR1 = 0; + /* + * On POWER9, we need to do a H_REGISTER_PROC_TBL hcall + * to inform the hypervisor that we wish to use the HPT. + */ + if (cpu_has_feature(CPU_FTR_ARCH_300)) + register_process_table(0, 0, 0); #ifdef CONFIG_FA_DUMP /* * If firmware assisted dump is active firmware preserves diff --git a/arch/powerpc/platforms/pseries/lpar.c b/arch/powerpc/platforms/pseries/lpar.c index 0587655..5b47026 100644 --- a/arch/powerpc/platforms/pseries/lpar.c +++ b/arch/powerpc/platforms/pseries/lpar.c @@ -609,15 +609,18 @@ static int __init disable_bulk_remove(char *str) __setup("bulk_remove=", disable_bulk_remove); -/* Actually only used for radix, so far */ static int pseries_lpar_register_process_table(unsigned long base, unsigned long page_size, unsigned long table_size) { long rc; - unsigned long flags = PROC_TABLE_NEW; + unsigned long flags = 0; + if (table_size) + flags |= PROC_TABLE_NEW; if (radix_enabled()) flags |= PROC_TABLE_RADIX | PROC_TABLE_GTSE; + else + flags |= PROC_TABLE_HPT_SLB; for (;;) { rc = plpar_hcall_norets(H_REGISTER_PROC_TBL, flags, base, page_size, table_size); @@ -643,6 +646,7 @@ void __init hpte_init_pseries(void) mmu_hash_ops.flush_hash_range = pSeries_lpar_flush_hash_range; mmu_hash_ops.hpte_clear_all = pseries_hpte_clear_all; mmu_hash_ops.hugepage_invalidate = pSeries_lpar_hugepage_invalidate; + register_process_table = pseries_lpar_register_process_table; } void radix_init_pseries(void)
On POWER9, since commit cc3d2940133d ("powerpc/64: Enable use of radix MMU under hypervisor on POWER9", 2017-01-30), we set both the radix and HPT bits in the client-architecture-support (CAS) vector, which tells the hypervisor that we can do either radix or HPT. According to PAPR, if we use this combination we are promising to do a H_REGISTER_PROC_TBL hcall later on to let the hypervisor know whether we are doing radix or HPT. We currently do this call if we are doing radix but not if we are doing HPT. If the hypervisor is able to support both radix and HPT guests, it would be entitled to defer allocation of the HPT until the H_REGISTER_PROC_TBL call, and to fail any attempts to create HPTEs until the H_REGISTER_PROC_TBL call. Thus we need to do a H_REGISTER_PROC_TBL call when we are doing HPT; otherwise we may crash at boot time. This adds the code to call H_REGISTER_PROC_TBL in this case, before we attempt to create any HPT entries using H_ENTER. Fixes: cc3d2940133d ("powerpc/64: Enable use of radix MMU under hypervisor on POWER9") Signed-off-by: Paul Mackerras <paulus@ozlabs.org> --- This needs to go in after the topic/ppc-kvm branch. arch/powerpc/mm/hash_utils_64.c | 6 ++++++ arch/powerpc/platforms/pseries/lpar.c | 8 ++++++-- 2 files changed, 12 insertions(+), 2 deletions(-)