riscv: some svpbmt fixes
v2 cover riscv: some svpbmt fixes
Some additionals comments and notes from autobuilders received after the series got applied, warranted some changes.
So this is a small collection of cleanups for the svpbmt v10 series.
changes in v2:
- add Guo's Review
- add patch dropping the use of function pointers in code that can run before relocation
Heiko Stuebner (5):
riscv: drop cpufeature_apply_feature tracking variable
riscv: Improve description for RISCV_ISA_SVPBMT Kconfig symbol
riscv: make patch-function pointer more generic in
cpu_manufacturer_info struct
riscv: fix dependency for t-head errata
riscv: remove usage of function-pointers from cpufeatures and t-head
errata
arch/riscv/Kconfig | 9 ++++++-- arch/riscv/Kconfig.erratas | 1 + arch/riscv/errata/thead/errata.c | 38 ++++++++++---------------------- arch/riscv/kernel/alternative.c | 18 +++++++-------- arch/riscv/kernel/cpufeature.c | 37 +++++++++---------------------- 5 files changed, 40 insertions(+), 63 deletions(-)
Signature · 1 line
2.35.1
PATCH v2 1/5 riscv: drop cpufeature_apply_feature tracking variable
The variable was tracking which feature patches got applied but that information was never actually used - and thus resulted in a warning as well.
Drop the variable.
--- arch/riscv/kernel/cpufeature.c | 5 +---- 1 file changed, 1 insertion(+), 4 deletions(-)
arch/riscv/kernel/cpufeature.c +1 -4
@@ -294,7 +294,6 @@ void __init_or_module riscv_cpufeature_patch_func(struct alt_entry *begin, unsigned int stage) {u32 cpu_req_feature =cpufeature_probe (stage); -u32 cpu_apply_feature =0 ;struct alt_entry *alt;u32 tmp;
@@ -308,10 +307,8 @@ void __init_or_module riscv_cpufeature_patch_func(struct alt_entry *begin, } tmp = (1U << alt->errata_id); -if (cpu_req_feature & tmp) { +if (cpu_req_feature & tmp)patch_text_nosync (alt->old_ptr, alt->alt_ptr, alt->alt_len); - cpu_apply_feature |= tmp; - } } }#endif
Signature · 1 line
2.35.1
Re: [PATCH v2 1/5] riscv: drop cpufeature_apply_feature tracking variable
Safe modification.
Quoted 40 lines · Heiko Stuebner
The variable was tracking which feature patches got applied but that information was never actually used - and thus resulted in a warning as well.
Drop the variable.
Reported-by kernel test robot <[email protected]>Signed-off-by Heiko Stuebner <[email protected]>--- arch/riscv/kernel/cpufeature.c | 5 +---- 1 file changed, 1 insertion(+), 4 deletions(-)arch/riscv/kernel/cpufeature.c +1
-5@@ -294,7 +294,6 @@ void __init_or_module riscv_cpufeature_patch_func(struct alt_entry *begin, unsigned int stage) {u32 cpu_req_feature =cpufeature_probe (stage); -u32 cpu_apply_feature =0 ;struct alt_entry *alt;u32 tmp;@@ -308,10 +307,8 @@ void __init_or_module riscv_cpufeature_patch_func(struct alt_entry *begin, } tmp = (1U << alt->errata_id); -if (cpu_req_feature & tmp) { +if (cpu_req_feature & tmp)patch_text_nosync (alt->old_ptr, alt->alt_ptr, alt->alt_len); - cpu_apply_feature |= tmp; - } } }#endif --2.35.1
Signature · 4 lines
Best Regards Guo Ren ML: https://lore.kernel.org/linux-csky/
PATCH v2 2/5 riscv: Improve description for RISCV_ISA_SVPBMT Kconfig symbol
In reply to Heiko StuebnerThis improves the symbol's description to make it easier for people to understand what it is about.
--- arch/riscv/Kconfig | 9 +++++++-- 1 file changed, 7 insertions(+), 2 deletions(-)
arch/riscv/Kconfig +7 -2
@@ -356,8 +356,13 @@ config RISCV_ISA_SVPBMT select RISCV_ALTERNATIVE default y help - Adds support to dynamically detect the presence of the SVPBMT extension - (Supervisor-mode: page-based memory types) and enable its usage. + Adds support to dynamically detect the presence of the SVPBMT + ISA-extension (Supervisor-mode: page-based memory types) and + enable its usage. + + The memory type for a page contains a combination of attributes + that indicate the cacheability, idempotency, and ordering + properties for access to that page. The SVPBMT extension is only available on 64Bit cpus.
Signature · 1 line
2.35.1
Re: [PATCH v2 2/5] riscv: Improve description for RISCV_ISA_SVPBMT Kconfig symbol
Quoted 34 lines · Heiko Stuebner
This improves the symbol's description to make it easier for people to understand what it is about.
Suggested-by Christoph Hellwig <[email protected]>Suggested-by Philipp Tomsich <[email protected]>Signed-off-by Heiko Stuebner <[email protected]>--- arch/riscv/Kconfig | 9 +++++++-- 1 file changed, 7 insertions(+), 2 deletions(-)arch/riscv/Kconfig +7
-3@@ -356,8 +356,13 @@ config RISCV_ISA_SVPBMT select RISCV_ALTERNATIVE default y help - Adds support to dynamically detect the presence of the SVPBMT extension - (Supervisor-mode: page-based memory types) and enable its usage. + Adds support to dynamically detect the presence of the SVPBMT + ISA-extension (Supervisor-mode: page-based memory types) and + enable its usage. + + The memory type for a page contains a combination of attributes + that indicate the cacheability, idempotency, and ordering + properties for access to that page. The SVPBMT extension is only available on 64Bit cpus. --2.35.1
Signature · 4 lines
Best Regards Guo Ren ML: https://lore.kernel.org/linux-csky/
PATCH v2 3/5 riscv: make patch-function pointer more generic in cpu_manufacturer_info struct
In reply to Heiko StuebnerDuring review the naming of the function-pointer was called confusing as the vendor id is just one of three inputs for the patching and indeed it serves no real purpose, as with recent changes the function pointer is not a static global element anymore, so drop the "vendor_" prefix.
--- arch/riscv/kernel/alternative.c | 18 +++++++++--------- 1 file changed, 9 insertions(+), 9 deletions(-)
arch/riscv/kernel/alternative.c +9 -9
@@ -20,7 +20,7 @@ struct cpu_manufacturer_info_t {unsigned long vendor_id;unsigned long arch_id;unsigned long imp_id; -void (*vendor_patch_func)(struct alt_entry *begin,struct alt_entry *end, +void (*patch_func)(struct alt_entry *begin,struct alt_entry *end,unsigned long archid,unsigned long impid,unsigned int stage); };
@@ -40,16 +40,16 @@ static void __init_or_module riscv_fill_cpu_mfr_info(struct cpu_manufacturer_infswitch (cpu_mfr_info->vendor_id) {#ifdef CONFIG_ERRATA_SIFIVE case SIFIVE_VENDOR_ID : - cpu_mfr_info->vendor_patch_func = sifive_errata_patch_func; + cpu_mfr_info->patch_func = sifive_errata_patch_func;break ;#endif #ifdef CONFIG_ERRATA_THEAD case THEAD_VENDOR_ID : - cpu_mfr_info->vendor_patch_func = thead_errata_patch_func; + cpu_mfr_info->patch_func = thead_errata_patch_func;break ;#endif default : - cpu_mfr_info->vendor_patch_func =NULL ; + cpu_mfr_info->patch_func =NULL ; } }
@@ -68,13 +68,13 @@ static void __init_or_module _apply_alternatives(struct alt_entry *begin,riscv_cpufeature_patch_func (begin, end, stage); -if (!cpu_mfr_info.vendor_patch_func) +if (!cpu_mfr_info.patch_func)return ; - cpu_mfr_info.vendor_patch_func (begin, end, - cpu_mfr_info.arch_id, - cpu_mfr_info.imp_id, - stage); + cpu_mfr_info.patch_func (begin, end, + cpu_mfr_info.arch_id, + cpu_mfr_info.imp_id, + stage); }void __init apply_boot_alternatives (void )
Signature · 1 line
2.35.1
PATCH v2 4/5 riscv: fix dependency for t-head errata
In reply to Heiko Stuebneralternatives only work correctly on non-xip-kernels and while the selected alternative-symbol has the correct dependency the symbol selecting it also needs that dependency.
So add the missing dependency to the T-Head errata Kconfig symbol.
--- arch/riscv/Kconfig.erratas | 1 + 1 file changed, 1 insertion(+)
arch/riscv/Kconfig.erratas +1 -0
@@ -35,6 +35,7 @@ config ERRATA_SIFIVE_CIP_1200 config ERRATA_THEAD bool "T-HEAD errata" + depends on !XIP_KERNEL select RISCV_ALTERNATIVE help All T-HEAD errata Kconfig depend on this Kconfig. Disabling
Signature · 1 line
2.35.1
Re: [PATCH v2 4/5] riscv: fix dependency for t-head errata
Em... This is a limitation to our scenario but seems it okay for current t-thead SoC products.
Quoted 29 lines · Heiko Stuebner
alternatives only work correctly on non-xip-kernels and while the selected alternative-symbol has the correct dependency the symbol selecting it also needs that dependency.
So add the missing dependency to the T-Head errata
Kconfigsymbol.Reported-by kernel test robot <[email protected]>Reviewed-by Guo Ren <[email protected]>Signed-off-by Heiko Stuebner <[email protected]>--- arch/riscv/Kconfig.erratas | 1 + 1 file changed, 1 insertion(+)arch/riscv/Kconfig.erratas +1
-1@@ -35,6 +35,7 @@ config ERRATA_SIFIVE_CIP_1200 config ERRATA_THEAD bool "T-HEAD errata" + depends on !XIP_KERNEL select RISCV_ALTERNATIVE help All T-HEAD errata Kconfig depend on this Kconfig. Disabling --2.35.1
Signature · 4 lines
Best Regards Guo Ren ML: https://lore.kernel.org/linux-csky/
PATCH v2 5/5 riscv: remove usage of function-pointers from cpufeatures and t-head errata
In reply to Heiko StuebnerHaving a list of alternatives to check with a per-entry function pointer to a check function is nice style-wise. But in case of early-alternatives it can clash with the non-relocated kernel and the function pointer in the list pointing to a completely wrong location.
This isn't an issue with one or two list entries, as in that case the compiler seems to unroll the loop and even usage of the list structure and then only does relative jumps into the check functions based on this.
When adding a third entry to either list though, the issue that was hiding there from the beginning is triggered resulting a jump to a memory address that isn't part of the kernel at all.
The list of features/erratas only contained an unused name and the pointer to the check function, so an easy solution for the problem is to just unroll the loop in code, dismantle the whole list structure and just call the relevant check functions one by one ourself.
For the T-Head errata this includes moving the stage-check inside the check functions.
The issue is only relevant for things that might be called for early- alternatives (T-Head and possible future main extensions), so the SiFive erratas were not affected from the beginning, as they got an early return for early-alternatives in the original patchset.
--- arch/riscv/errata/thead/errata.c | 38 ++++++++++---------------------- arch/riscv/kernel/cpufeature.c | 32 +++++++++------------------ 2 files changed, 22 insertions(+), 48 deletions(-)
arch/riscv/errata/thead/errata.c +12 -26
@@ -14,40 +14,26 @@#include <asm/patch.h> #include <asm/vendorid_list.h> -struct errata_info { -char name[ERRATA_STRING_LENGTH_MAX ]; -bool (*check_func)(unsigned long arch_id,unsigned long impid); -unsigned int stage; -}; - -static bool errata_mt_check_func (unsigned long arch_id,unsigned long impid) +static bool errata_probe_pbmt (unsigned int stage, +unsigned long arch_id,unsigned long impid) {if (arch_id !=0 || impid !=0 )return false; -return true; -} -static const struct errata_info errata_list[ERRATA_THEAD_NUMBER ] = { - { - .name ="memory-types" , - .stage =RISCV_ALTERNATIVES_EARLY_BOOT , - .check_func = errata_mt_check_func - }, -}; +if (stage ==RISCV_ALTERNATIVES_EARLY_BOOT || + stage ==RISCV_ALTERNATIVES_MODULE ) +return true; + +return false; +} -static u32 thead_errata_probe (unsigned int stage,unsigned long archid,unsigned long impid) +static u32 thead_errata_probe (unsigned int stage, +unsigned long archid,unsigned long impid) { -const struct errata_info *info;u32 cpu_req_errata =0 ; -int idx; - -for (idx =0 ; idx <ERRATA_THEAD_NUMBER ; idx++) { - info = &errata_list[idx]; -if ((stage ==RISCV_ALTERNATIVES_MODULE || - info->stage == stage) && info->check_func (archid, impid)) - cpu_req_errata |= (1U << idx); - } +if (errata_probe_pbmt (stage, archid, impid)) + cpu_req_errata |= (1U <<ERRATA_THEAD_PBMT );return cpu_req_errata; }
arch/riscv/kernel/cpufeature.c +10 -22
@@ -246,12 +246,7 @@ void __init riscv_fill_hwcap(void) }#ifdef CONFIG_RISCV_ALTERNATIVE -struct cpufeature_info { -char name[ERRATA_STRING_LENGTH_MAX ]; -bool (*check_func)(unsigned int stage); -}; - -static bool __init_or_module cpufeature_svpbmt_check_func(unsigned int stage) +static bool __init_or_module cpufeature_probe_svpbmt(unsigned int stage) {#ifdef CONFIG_RISCV_ISA_SVPBMT switch (stage) {
@@ -265,26 +260,19 @@ static bool __init_or_module cpufeature_svpbmt_check_func(unsigned int stage)return false; } -static const struct cpufeature_info __initdata_or_module -cpufeature_list[CPUFEATURE_NUMBER] = { - { - .name ="svpbmt" , - .check_func = cpufeature_svpbmt_check_func - }, -}; - +/* +* Probe presence of individual extensions. +* +* This code may also be executed before kernel relocation, so we cannot use +* addresses generated by the address-of operator as they won't be valid in +* this context. +*/ static u32 __init_or_module cpufeature_probe(unsigned int stage) { -const struct cpufeature_info *info;u32 cpu_req_feature =0 ; -int idx; - -for (idx =0 ; idx <CPUFEATURE_NUMBER ; idx++) { - info = &cpufeature_list[idx]; -if (info->check_func (stage)) - cpu_req_feature |= (1U << idx); - } +if (cpufeature_probe_svpbmt (stage)) + cpu_req_feature |= (1U <<CPUFEATURE_SVPBMT );return cpu_req_feature; }
Signature · 1 line
2.35.1
Re: [PATCH v2 5/5] riscv: remove usage of function-pointers from cpufeatures and t-head errata
Quoted 94 lines · Heiko Stuebner
Having a list of alternatives to check with a per-entry function pointer to a check function is nice style-wise. But in case of early-alternatives it can clash with the non-relocated kernel and the function pointer in the list pointing to a completely wrong location.
This isn't an issue with one or two list entries, as in that case the compiler seems to unroll the loop and even usage of the list structure and then only does relative jumps into the check functions based on this.
When adding a third entry to either list though, the issue that was hiding there from the beginning is triggered resulting a jump to a memory address that isn't part of the kernel at all.
The list of features/erratas only contained an unused name and the pointer to the check function, so an easy solution for the problem is to just unroll the loop in code, dismantle the whole list structure and just call the relevant check functions one by one ourself.
For the T-Head errata this includes moving the stage-check inside the check functions.
The issue is only relevant for things that might be called for early- alternatives (T-Head and possible future main extensions), so the SiFive erratas were not affected from the beginning, as they got an early return for early-alternatives in the original patchset.
Signed-off-by Heiko Stuebner <[email protected]>--- arch/riscv/errata/thead/errata.c | 38 ++++++++++---------------------- arch/riscv/kernel/cpufeature.c | 32 +++++++++------------------ 2 files changed, 22 insertions(+), 48 deletions(-)arch/riscv/errata/thead/errata.c +12
-26@@ -14,40 +14,26 @@#include <asm/patch.h> #include <asm/vendorid_list.h> -struct errata_info { -char name[ERRATA_STRING_LENGTH_MAX ]; -bool (*check_func)(unsigned long arch_id,unsigned long impid); -unsigned int stage; -}; - -static bool errata_mt_check_func (unsigned long arch_id,unsigned long impid) +static bool errata_probe_pbmt (unsigned int stage, +unsigned long arch_id,unsigned long impid) {if (arch_id !=0 || impid !=0 )return false; -return true; -} -static const struct errata_info errata_list[ERRATA_THEAD_NUMBER ] = { - { - .name ="memory-types" , - .stage =RISCV_ALTERNATIVES_EARLY_BOOT , - .check_func = errata_mt_check_func - }, -}; +if (stage ==RISCV_ALTERNATIVES_EARLY_BOOT || + stage ==RISCV_ALTERNATIVES_MODULE ) +return true; + +return false; +} -static u32 thead_errata_probe (unsigned int stage,unsigned long archid,unsigned long impid) +static u32 thead_errata_probe (unsigned int stage, +unsigned long archid,unsigned long impid) { -const struct errata_info *info;u32 cpu_req_errata =0 ; -int idx; - -for (idx =0 ; idx <ERRATA_THEAD_NUMBER ; idx++) { - info = &errata_list[idx]; -if ((stage ==RISCV_ALTERNATIVES_MODULE || - info->stage == stage) && info->check_func (archid, impid)) - cpu_req_errata |= (1U << idx); - } +if (errata_probe_pbmt (stage, archid, impid)) + cpu_req_errata |= (1U <<ERRATA_THEAD_PBMT );return cpu_req_errata; }arch/riscv/kernel/cpufeature.c +0
-0
arch/riscv/kernel/cpufeature.c void __init riscv_fill_hwcap(void) · line 249
Quoted 10 lines
arch/riscv/kernel/cpufeature.c +1
-6@@ -246,12 +246,7 @@ void __init riscv_fill_hwcap(void) }#ifdef CONFIG_RISCV_ALTERNATIVE -struct cpufeature_info { -char name[ERRATA_STRING_LENGTH_MAX ]; -bool (*check_func)(unsigned int stage); -}; - -static bool __init_or_module cpufeature_svpbmt_check_func(unsigned int stage) +static bool __init_or_module cpufeature_probe_svpbmt(unsigned int stage)
Now that this function isn't used as a function pointer anymore, it doesn't need to be specific to SVPBMT. I think the logic here is the same for ZICBOM. Does it make sense to move it to the calling function?
With the conflicts between this and the CMO series manually fixed:
Regards,
Samuel
Quoted 40 lines
arch/riscv/kernel/cpufeature.c +9
-16{#ifdef CONFIG_RISCV_ISA_SVPBMT switch (stage) {@@ -265,26 +260,19 @@ static bool __init_or_module cpufeature_svpbmt_check_func(unsigned int stage)return false; } -static const struct cpufeature_info __initdata_or_module -cpufeature_list[CPUFEATURE_NUMBER] = { - { - .name ="svpbmt" , - .check_func = cpufeature_svpbmt_check_func - }, -}; - +/* +* Probe presence of individual extensions. +* +* This code may also be executed before kernel relocation, so we cannot use +* addresses generated by the address-of operator as they won't be valid in +* this context. +*/ static u32 __init_or_module cpufeature_probe(unsigned int stage) { -const struct cpufeature_info *info;u32 cpu_req_feature =0 ; -int idx; - -for (idx =0 ; idx <CPUFEATURE_NUMBER ; idx++) { - info = &cpufeature_list[idx]; -if (info->check_func (stage)) - cpu_req_feature |= (1U << idx); - } +if (cpufeature_probe_svpbmt (stage)) + cpu_req_feature |= (1U <<CPUFEATURE_SVPBMT );return cpu_req_feature; }
Re: [PATCH v2 5/5] riscv: remove usage of function-pointers from cpufeatures and t-head errata
Hi,
Am Sonntag, 29. Mai 2022, 03:27:48 CEST schrieb Samuel Holland:
Quoted 28 lines · Heiko Stuebner
Quoted 27 lines
Having a list of alternatives to check with a per-entry function pointer to a check function is nice style-wise. But in case of early-alternatives it can clash with the non-relocated kernel and the function pointer in the list pointing to a completely wrong location.
This isn't an issue with one or two list entries, as in that case the compiler seems to unroll the loop and even usage of the list structure and then only does relative jumps into the check functions based on this.
When adding a third entry to either list though, the issue that was hiding there from the beginning is triggered resulting a jump to a memory address that isn't part of the kernel at all.
The list of features/erratas only contained an unused name and the pointer to the check function, so an easy solution for the problem is to just unroll the loop in code, dismantle the whole list structure and just call the relevant check functions one by one ourself.
For the T-Head errata this includes moving the stage-check inside the check functions.
The issue is only relevant for things that might be called for early- alternatives (T-Head and possible future main extensions), so the SiFive erratas were not affected from the beginning, as they got an early return for early-alternatives in the original patchset.
Signed-off-by Heiko Stuebner <[email protected]>
[...]
Quoted 27 lines
Quoted 27 lines
arch/riscv/errata/thead/errata.c +4
-10-static u32 thead_errata_probe (unsigned int stage,unsigned long archid,unsigned long impid) +static u32 thead_errata_probe (unsigned int stage, +unsigned long archid,unsigned long impid) { -const struct errata_info *info;u32 cpu_req_errata =0 ; -int idx; - -for (idx =0 ; idx <ERRATA_THEAD_NUMBER ; idx++) { - info = &errata_list[idx]; -if ((stage ==RISCV_ALTERNATIVES_MODULE || - info->stage == stage) && info->check_func (archid, impid)) - cpu_req_errata |= (1U << idx); - } +if (errata_probe_pbmt (stage, archid, impid)) + cpu_req_errata |= (1U <<ERRATA_THEAD_PBMT );return cpu_req_errata; }arch/riscv/kernel/cpufeature.c +0
-0@@ -246,12 +246,7 @@ void __init riscv_fill_hwcap(void) }
Quoted 12 lines
arch/riscv/kernel/cpufeature.c void __init riscv_fill_hwcap(void) · line 249
Quoted 8 lines
arch/riscv/kernel/cpufeature.c +1
-6#ifdef CONFIG_RISCV_ALTERNATIVE -struct cpufeature_info { -char name[ERRATA_STRING_LENGTH_MAX ]; -bool (*check_func)(unsigned int stage); -}; - -static bool __init_or_module cpufeature_svpbmt_check_func(unsigned int stage) +static bool __init_or_module cpufeature_probe_svpbmt(unsigned int stage)Now that this function isn't used as a function pointer anymore, it doesn't need to be specific to SVPBMT. I think the logic here is the same for ZICBOM. Does it make sense to move it to the calling function?
we don't know how other extensions need to probe though, so for example in my yet-to-send zicbom-v3 the probe function itself looks like
static u32 __init_or_module cpufeature_probe(unsigned int stage)
{
u32 cpu_req_feature = 0;
if (cpufeature_probe_svpbmt(stage))
cpu_req_feature |= (1U << CPUFEATURE_SVPBMT);
if (cpufeature_probe_cmo(stage))
cpu_req_feature |= (1U << CPUFEATURE_CMO);
return cpu_req_feature;
}As this might get longer in the future, I actually do like the actual checks being separate for readability.
But I'll just yield to the majority opinion ;-)
Heiko
Quoted 45 lines
With the conflicts between this and the CMO series manually fixed:
Tested-by Samuel Holland <[email protected]>Regards,
SamuelQuoted 37 lines
arch/riscv/kernel/cpufeature.c +9
-16{#ifdef CONFIG_RISCV_ISA_SVPBMT switch (stage) {@@ -265,26 +260,19 @@ static bool __init_or_module cpufeature_svpbmt_check_func(unsigned int stage)return false; } -static const struct cpufeature_info __initdata_or_module -cpufeature_list[CPUFEATURE_NUMBER] = { - { - .name ="svpbmt" , - .check_func = cpufeature_svpbmt_check_func - }, -}; - +/* +* Probe presence of individual extensions. +* +* This code may also be executed before kernel relocation, so we cannot use +* addresses generated by the address-of operator as they won't be valid in +* this context. +*/ static u32 __init_or_module cpufeature_probe(unsigned int stage) { -const struct cpufeature_info *info;u32 cpu_req_feature =0 ; -int idx; - -for (idx =0 ; idx <CPUFEATURE_NUMBER ; idx++) { - info = &cpufeature_list[idx]; -if (info->check_func (stage)) - cpu_req_feature |= (1U << idx); - } +if (cpufeature_probe_svpbmt (stage)) + cpu_req_feature |= (1U <<CPUFEATURE_SVPBMT );
Quoted 5 lines
arch/riscv/kernel/cpufeature.c static bool __init_or_module cpufeature_svpbmt_check_func(unsigned int stage) · line 278
Quoted 3 lines
arch/riscv/kernel/cpufeature.c +0
-0return cpu_req_feature; }
_______________________________________________
linux-riscv mailing list
[email protected]
http://lists.infradead.org/mailman/listinfo/linux-riscv
Palmer Dabbelt
maintainerQuoted 19 lines · On Thu, 26 May 2022 13:56:41 PDT (-0700), [email protected] wrote:
Some additionals comments and notes from autobuilders received after the series got applied, warranted some changes.
So this is a small collection of cleanups for the svpbmt v10 series.
changes in v2:
- add Guo's Review
- add patch dropping the use of function pointers in code that can run before relocationHeiko Stuebner (5):
riscv: dropcpufeature_apply_featuretracking variable
riscv: Improve description forRISCV_ISA_SVPBMTKconfigsymbol
riscv: make patch-function pointer more generic in
cpu_manufacturer_infostruct
riscv: fix dependency for t-head errata
riscv: remove usage of function-pointers from cpufeatures and t-head
errata
Quoted 6 lines
arch/riscv/Kconfig | 9 ++++++-- arch/riscv/Kconfig.erratas | 1 + arch/riscv/errata/thead/errata.c | 38 ++++++++++---------------------- arch/riscv/kernel/alternative.c | 18 +++++++-------- arch/riscv/kernel/cpufeature.c | 37 +++++++++---------------------- 5 files changed, 40 insertions(+), 63 deletions(-)
IMO only three of these are actually fixes, the rest are cleanups. I've got ahead and put everything on a branch, with
riscv: Improve description for RISCV_ISA_SVPBMT Kconfig symbol
riscv: drop cpufeature_apply_feature tracking variable
riscv: fix dependency for t-head errata
first. Those are on fixes, the whole thing is in for-next.
Thanks!
Hi Palmer,
Am Freitag, 17. Juni 2022, 01:03:32 CEST schrieb Palmer Dabbelt:
Quoted 27 lines · On Thu, 26 May 2022 13:56:41 PDT (-0700), [email protected] wrote:
Quoted 19 lines
Some additionals comments and notes from autobuilders received after the series got applied, warranted some changes.
So this is a small collection of cleanups for the svpbmt v10 series.
changes in v2:
- add Guo's Review
- add patch dropping the use of function pointers in code that can run before relocationHeiko Stuebner (5):
riscv: dropcpufeature_apply_featuretracking variable
riscv: Improve description forRISCV_ISA_SVPBMTKconfigsymbol
riscv: make patch-function pointer more generic in
cpu_manufacturer_infostruct
riscv: fix dependency for t-head errata
riscv: remove usage of function-pointers from cpufeatures and t-head
errataQuoted 6 lines
arch/riscv/Kconfig | 9 ++++++-- arch/riscv/Kconfig.erratas | 1 + arch/riscv/errata/thead/errata.c | 38 ++++++++++---------------------- arch/riscv/kernel/alternative.c | 18 +++++++-------- arch/riscv/kernel/cpufeature.c | 37 +++++++++---------------------- 5 files changed, 40 insertions(+), 63 deletions(-)
Quoted 8 lines
IMO only three of these are actually fixes, the rest are cleanups. I've got ahead and put everything on a branch, with
riscv: Improve description for
RISCV_ISA_SVPBMTKconfigsymbol
riscv: dropcpufeature_apply_featuretracking variable
riscv: fix dependency for t-head erratafirst. Those are on fixes, the whole thing is in for-next.
thanks a lot.
For the fixes I think Nathan's "riscv: Fix ALT_THEAD_PMA's asm parameters" [0] would also be quite important.
Heiko