target/xtensa: tidy xtensa memory management
PATCH 1/2 target/xtensa: wrap MMU and MPU state into structures
In reply to Max FilippovMake separation of alternative xtensa memory management options state explicit.
--- target/xtensa/cpu.h | 18 +++++++++++++---- target/xtensa/mmu_helper.c | 40 +++++++++++++++++++------------------- 2 files changed, 34 insertions(+), 24 deletions(-)
target/xtensa/cpu.h +14 -4
@@ -326,11 +326,21 @@ typedef struct xtensa_tlb {unsigned nrefillentries; } xtensa_tlb; +typedef struct XtensaMMU { +xtensa_tlb_entry itlb[7 ][MAX_TLB_WAY_SIZE ]; +xtensa_tlb_entry dtlb[10 ][MAX_TLB_WAY_SIZE ]; +unsigned autorefill_idx; +}XtensaMMU ; +typedef struct xtensa_mpu_entry {uint32_t vaddr;uint32_t attr; }xtensa_mpu_entry ; +typedef struct XtensaMPU { +xtensa_mpu_entry fg[MAX_MPU_FOREGROUND_SEGMENTS ]; +}XtensaMPU ; + typedef struct XtensaGdbReg { int targno; unsigned flags;
@@ -526,10 +536,10 @@ struct CPUArchState {uint32_t exclusive_val;#ifndef CONFIG_USER_ONLY -xtensa_tlb_entry itlb[7 ][MAX_TLB_WAY_SIZE ]; -xtensa_tlb_entry dtlb[10 ][MAX_TLB_WAY_SIZE ]; -xtensa_mpu_entry mpu_fg[MAX_MPU_FOREGROUND_SEGMENTS ]; -unsigned autorefill_idx; +union { +XtensaMMU mmu; +XtensaMPU mpu; + };bool runstall;AddressSpace *address_space_er;MemoryRegion *system_er;
target/xtensa/mmu_helper.c +20 -20
@@ -250,8 +250,8 @@ static xtensa_tlb_entry *xtensa_tlb_get_entry(CPUXtensaState *env, bool dtlb,assert (wi < tlb->nways && ei < tlb->way_size[wi]);return dtlb ? - env->dtlb[wi] + ei : - env->itlb[wi] + ei; + env->mmu.dtlb[wi] + ei : + env->mmu.itlb[wi] + ei; } static xtensa_tlb_entry *get_tlb_entry(CPUXtensaState *env,
@@ -411,11 +411,11 @@ void reset_mmu(CPUXtensaState *env) env->sregs[RASID ] =0x04030201 ; env->sregs[ITLBCFG ] =0 ; env->sregs[DTLBCFG ] =0 ; - env->autorefill_idx =0 ; -reset_tlb_mmu_all_ways (env, &env->config->itlb, env->itlb); -reset_tlb_mmu_all_ways (env, &env->config->dtlb, env->dtlb); -reset_tlb_mmu_ways56 (env, &env->config->itlb, env->itlb); -reset_tlb_mmu_ways56 (env, &env->config->dtlb, env->dtlb); + env->mmu.autorefill_idx =0 ; +reset_tlb_mmu_all_ways (env, &env->config->itlb, env->mmu.itlb); +reset_tlb_mmu_all_ways (env, &env->config->dtlb, env->mmu.dtlb); +reset_tlb_mmu_ways56 (env, &env->config->itlb, env->mmu.itlb); +reset_tlb_mmu_ways56 (env, &env->config->dtlb, env->mmu.dtlb); }else if (xtensa_option_enabled (env->config,XTENSA_OPTION_MPU )) {unsigned i;
@@ -430,8 +430,8 @@ void reset_mmu(CPUXtensaState *env) } }else { env->sregs[CACHEATTR ] =0x22222222 ; -reset_tlb_region_way0 (env, env->itlb); -reset_tlb_region_way0 (env, env->dtlb); +reset_tlb_region_way0 (env, env->mmu.itlb); +reset_tlb_region_way0 (env, env->mmu.dtlb); } }
@@ -462,7 +462,7 @@ static int xtensa_tlb_lookup(const CPUXtensaState *env,const xtensa_tlb *tlb = dtlb ? &env->config->dtlb : &env->config->itlb;const xtensa_tlb_entry (*entry)[MAX_TLB_WAY_SIZE ] = dtlb ? - env->dtlb : env->itlb; + env->mmu.dtlb : env->mmu.itlb;int nhits =0 ;unsigned wi;
@@ -821,7 +821,7 @@ static int get_physical_addr_mmu(CPUXtensaState *env, bool update_tlb,split_tlb_entry_spec_way (env, vaddr, dtlb, &vpn, wi, &ei);if (update_tlb) { - wi = ++env->autorefill_idx &0x3 ; + wi = ++env->mmu.autorefill_idx &0x3 ;xtensa_tlb_set_entry (env, dtlb, wi, ei, vpn, pte); env->sregs[EXCVADDR ] = vaddr; qemu_log_mask(CPU_LOG_MMU, "%s: autorefill(%08x): %08x -> %08x\n",
@@ -957,8 +957,8 @@ void HELPER(wptlb)(CPUXtensaState *env, uint32_t p, uint32_t v)unsigned segment = p &XTENSA_MPU_SEGMENT_MASK ;if (segment < env->config->n_mpu_fg_segments) { - env->mpu_fg[segment].vaddr = v & -env->config->mpu_align; - env->mpu_fg[segment].attr = p &XTENSA_MPU_ATTR_MASK ; + env->mpu.fg[segment].vaddr = v & -env->config->mpu_align; + env->mpu.fg[segment].attr = p &XTENSA_MPU_ATTR_MASK ; env->sregs[MPUENB ] =deposit32 (env->sregs[MPUENB ], segment,1 , v);tlb_flush (env_cpu (env)); }
@@ -969,7 +969,7 @@ uint32_t HELPER(rptlb0)(CPUXtensaState *env, uint32_t s)unsigned segment = s &XTENSA_MPU_SEGMENT_MASK ;if (segment < env->config->n_mpu_fg_segments) { -return env->mpu_fg[segment].vaddr | +return env->mpu.fg[segment].vaddr |extract32 (env->sregs[MPUENB ], segment,1 ); }else {return 0 ;
@@ -981,7 +981,7 @@ uint32_t HELPER(rptlb1)(CPUXtensaState *env, uint32_t s)unsigned segment = s &XTENSA_MPU_SEGMENT_MASK ;if (segment < env->config->n_mpu_fg_segments) { -return env->mpu_fg[segment].attr; +return env->mpu.fg[segment].attr; }else {return 0 ; }
@@ -993,13 +993,13 @@ uint32_t HELPER(pptlb)(CPUXtensaState *env, uint32_t v)unsigned segment =XTENSA_MPU_PROBE_B ;unsigned bg_segment; - nhits =xtensa_mpu_lookup (env->mpu_fg, env->config->n_mpu_fg_segments, + nhits =xtensa_mpu_lookup (env->mpu.fg, env->config->n_mpu_fg_segments, v, &segment);if (nhits >1 ) {HELPER (exception_cause_vaddr)(env, env->pc,LOAD_STORE_TLB_MULTI_HIT_CAUSE , v); }else if (nhits ==1 && (env->sregs[MPUENB ] & (1u << segment))) { -return env->mpu_fg[segment].attr | segment |XTENSA_MPU_PROBE_V ; +return env->mpu.fg[segment].attr | segment |XTENSA_MPU_PROBE_V ; }else { xtensa_mpu_lookup(env->config->mpu_bg, env->config->n_mpu_bg_segments,
@@ -1017,14 +1017,14 @@ static int get_physical_addr_mpu(CPUXtensaState *env,unsigned segment;uint32_t attr; - nhits =xtensa_mpu_lookup (env->mpu_fg, env->config->n_mpu_fg_segments, + nhits =xtensa_mpu_lookup (env->mpu.fg, env->config->n_mpu_fg_segments, vaddr, &segment);if (nhits >1 ) {return is_write <2 ?LOAD_STORE_TLB_MULTI_HIT_CAUSE :INST_TLB_MULTI_HIT_CAUSE ; }else if (nhits ==1 && (env->sregs[MPUENB ] & (1u << segment))) { - attr = env->mpu_fg[segment].attr; + attr = env->mpu.fg[segment].attr; }else { xtensa_mpu_lookup(env->config->mpu_bg, env->config->n_mpu_bg_segments,
@@ -1205,7 +1205,7 @@ void dump_mmu(CPUXtensaState *env)dump_tlb (env, true); }else if (xtensa_option_enabled (env->config,XTENSA_OPTION_MPU )) {qemu_printf ("Foreground map:\n" ); -dump_mpu (env, env->mpu_fg, env->config->n_mpu_fg_segments); +dump_mpu (env, env->mpu.fg, env->config->n_mpu_fg_segments);qemu_printf ("\nBackground map:\n" );dump_mpu (NULL , env->config->mpu_bg, env->config->n_mpu_bg_segments); }else {
Signature · 1 line
2.39.2
cover target/xtensa: tidy xtensa memory management
Hello,
this series separates xtensa MMU and MPU states and improves variable TLB way logic.
Max Filippov (2):
target/xtensa: wrap MMU and MPU state into structures target/xtensa: tidy TLB way variability logic
target/xtensa/cpu.h | 21 +++++++--- target/xtensa/mmu_helper.c | 74 ++++++++++++++---------------------- target/xtensa/overlay_tool.h | 15 +++++++- 3 files changed, 56 insertions(+), 54 deletions(-)
Signature · 1 line
2.39.2
PATCH 2/2 target/xtensa: tidy TLB way variability logic
Whether TLB ways 5 and 6 are variable is not a property of the TLB instance or a TLB entry instance, it's a property of the xtensa core configuration.
Remove 'varway56' field from the xtensa_tlb structure and remove 'variable' field from the xtensa_tlb_entry structure. Add 'tlb_variable_way' array to the XtensaConfig and use it instead of removed fields.
--- target/xtensa/cpu.h | 3 +-- target/xtensa/mmu_helper.c | 38 ++++++++++-------------------------- target/xtensa/overlay_tool.h | 15 ++++++++++++-- 3 files changed, 24 insertions(+), 32 deletions(-)
target/xtensa/cpu.h +1 -2
@@ -316,13 +316,11 @@ typedef struct xtensa_tlb_entry {uint32_t paddr;uint8_t asid;uint8_t attr; -bool variable; } xtensa_tlb_entry;typedef struct xtensa_tlb {unsigned nways;const unsigned way_size[10 ]; -bool varway56;unsigned nrefillentries; }xtensa_tlb ;
@@ -493,6 +491,7 @@ typedef struct XtensaConfig {xtensa_tlb itlb;xtensa_tlb dtlb; +bool tlb_variable_way[16 ];uint32_t mpu_align;unsigned n_mpu_fg_segments;
target/xtensa/mmu_helper.c +10 -28
@@ -105,23 +105,19 @@ static uint32_t xtensa_tlb_get_addr_mask(const CPUXtensaState *env, bool dtlb, uint32_t way) {if (xtensa_option_enabled (env->config,XTENSA_OPTION_MMU )) { -bool varway56 = dtlb ? - env->config->dtlb.varway56 : - env->config->itlb.varway56; -switch (way) {case 4 :return 0xfff00000 <<get_page_size (env, dtlb, way) *2 ;case 5 : -if (varway56) { +if (env->config->tlb_variable_way[5 ]) {return 0xf8000000 <<get_page_size (env, dtlb, way); }else {return 0xf8000000 ; }case 6 : -if (varway56) { +if (env->config->tlb_variable_way[6 ]) {return 0xf0000000 << (1 -get_page_size (env, dtlb, way)); }else {return 0xf0000000 ;
@@ -150,11 +146,8 @@ static uint32_t get_vpn_mask(const CPUXtensaState *env, bool dtlb, uint32_t way)return xtensa_tlb_get_addr_mask (env, dtlb, way) <<2 ; }else if (way <=6 ) {uint32_t mask =xtensa_tlb_get_addr_mask (env, dtlb, way); -bool varway56 = dtlb ? - env->config->dtlb.varway56 : - env->config->itlb.varway56; -if (varway56) { +if (env->config->tlb_variable_way[5 ]) {return mask << (way ==5 ?2 :3 ); }else {return mask <<1 ;
@@ -172,10 +165,6 @@ static void split_tlb_entry_spec_way(const CPUXtensaState *env, uint32_t v, bool dtlb, uint32_t *vpn, uint32_t wi, uint32_t *ei) { - bool varway56 = dtlb ? - env->config->dtlb.varway56 : - env->config->itlb.varway56; - if (!dtlb) { wi &= 7; }
@@ -195,7 +184,7 @@ static void split_tlb_entry_spec_way(const CPUXtensaState *env, uint32_t v,break ;case 5 : -if (varway56) { +if (env->config->tlb_variable_way[5 ]) {uint32_t eibase =27 +get_page_size (env, dtlb, wi); *ei = (v >> eibase) &0x3 ; }else {
@@ -204,7 +193,7 @@ static void split_tlb_entry_spec_way(const CPUXtensaState *env, uint32_t v,break ;case 6 : -if (varway56) { +if (env->config->tlb_variable_way[6 ]) {uint32_t eibase =29 -get_page_size (env, dtlb, wi); *ei = (v >> eibase) &0x7 ; }else {
@@ -290,7 +279,7 @@ static void xtensa_tlb_set_entry(CPUXtensaState *env, bool dtlb,xtensa_tlb_entry *entry =xtensa_tlb_get_entry (env, dtlb, wi, ei);if (xtensa_option_enabled (env->config,XTENSA_OPTION_MMU )) { -if (entry->variable) { +if (env->config->tlb_variable_way[wi]) {if (entry->asid) {tlb_flush_page (cs, entry->vaddr); }
@@ -338,29 +327,25 @@ static void reset_tlb_mmu_all_ways(CPUXtensaState *env,for (wi =0 ; wi < tlb->nways; ++wi) {for (ei =0 ; ei < tlb->way_size[wi]; ++ei) { entry[wi][ei].asid =0 ; - entry[wi][ei].variable = true; } } }static void reset_tlb_mmu_ways56 (CPUXtensaState *env, -const xtensa_tlb *tlb,xtensa_tlb_entry entry[][MAX_TLB_WAY_SIZE ]) { -if (!tlb->varway56) { +if (!env->config->tlb_variable_way[5 ]) {static const xtensa_tlb_entry way5[] = { { .vaddr =0xd0000000 , .paddr =0 , .asid =1 , .attr =7 , - .variable = false, }, { .vaddr =0xd8000000 , .paddr =0 , .asid =1 , .attr =3 , - .variable = false, } }; static const xtensa_tlb_entry way6[] = {
@@ -369,13 +354,11 @@ static void reset_tlb_mmu_ways56(CPUXtensaState *env, .paddr = 0xf0000000, .asid = 1, .attr = 7, - .variable = false, }, { .vaddr = 0xf0000000, .paddr = 0xf0000000, .asid = 1, .attr = 3, - .variable = false, } };memcpy (entry[5 ], way5,sizeof (way5));
@@ -401,7 +384,6 @@ static void reset_tlb_region_way0(CPUXtensaState *env, entry[0 ][ei].paddr = ei <<29 ; entry[0 ][ei].asid =1 ; entry[0 ][ei].attr =2 ; - entry[0 ][ei].variable = true; } }
@@ -414,8 +396,8 @@ void reset_mmu(CPUXtensaState *env) env->mmu.autorefill_idx =0 ;reset_tlb_mmu_all_ways (env, &env->config->itlb, env->mmu.itlb);reset_tlb_mmu_all_ways (env, &env->config->dtlb, env->mmu.dtlb); -reset_tlb_mmu_ways56 (env, &env->config->itlb, env->mmu.itlb); -reset_tlb_mmu_ways56 (env, &env->config->dtlb, env->mmu.dtlb); +reset_tlb_mmu_ways56 (env, env->mmu.itlb); +reset_tlb_mmu_ways56 (env, env->mmu.dtlb); }else if (xtensa_option_enabled (env->config,XTENSA_OPTION_MPU )) {unsigned i;
@@ -521,7 +503,7 @@ void HELPER(itlb)(CPUXtensaState *env, uint32_t v, uint32_t dtlb)if (xtensa_option_enabled (env->config,XTENSA_OPTION_MMU )) {uint32_t wi;xtensa_tlb_entry *entry =get_tlb_entry (env, v, dtlb, &wi); -if (entry && entry->variable && entry->asid) { +if (entry && env->config->tlb_variable_way[wi] && entry->asid) {tlb_flush_page (env_cpu (env), entry->vaddr); entry->asid =0 ; }
target/xtensa/overlay_tool.h +13 -2
@@ -351,7 +351,6 @@ (refill_way_size), (refill_way_size), \ 4, (way56) ? 4 : 2, (way56) ? 8 : 2, 1, 1, 1, \ }, \ - .varway56 = (way56), \ .nrefillentries = (refill_way_size) * 4, \ }
@@ -363,7 +362,19 @@#define TLB_SECTION \ .itlb = ITLB(XCHAL_HAVE_SPANNING_WAY), \ - .dtlb = DTLB(XCHAL_HAVE_SPANNING_WAY) + .dtlb = DTLB(XCHAL_HAVE_SPANNING_WAY), \ + .tlb_variable_way = { \ + true, \ + XCHAL_HAVE_PTP_MMU, \ + XCHAL_HAVE_PTP_MMU, \ + XCHAL_HAVE_PTP_MMU, \ + XCHAL_HAVE_PTP_MMU, \ + XCHAL_HAVE_PTP_MMU && XCHAL_HAVE_SPANNING_WAY, \ + XCHAL_HAVE_PTP_MMU && XCHAL_HAVE_SPANNING_WAY, \ + XCHAL_HAVE_PTP_MMU, \ + XCHAL_HAVE_PTP_MMU, \ + XCHAL_HAVE_PTP_MMU, \ + }#ifndef XCHAL_SYSROM0_PADDR #define XCHAL_SYSROM0_PADDR 0xfe000000
Signature · 1 line
2.39.2
Re: [PATCH 1/2] target/xtensa: wrap MMU and MPU state into structures
In reply to Max FilippovQuoted 38 lines · Max Filippov
Make separation of alternative xtensa memory management options state explicit.
Signed-off-by Max Filippov <[email protected]>--- target/xtensa/cpu.h | 18 +++++++++++++---- target/xtensa/mmu_helper.c | 40 +++++++++++++++++++------------------- 2 files changed, 34 insertions(+), 24 deletions(-)target/xtensa/cpu.h +10
-0@@ -326,11 +326,21 @@ typedef struct xtensa_tlb {unsigned nrefillentries; } xtensa_tlb; +typedef struct XtensaMMU { +xtensa_tlb_entry itlb[7 ][MAX_TLB_WAY_SIZE ]; +xtensa_tlb_entry dtlb[10 ][MAX_TLB_WAY_SIZE ]; +unsigned autorefill_idx; +}XtensaMMU ; +typedef struct xtensa_mpu_entry {uint32_t vaddr;uint32_t attr; }xtensa_mpu_entry ; +typedef struct XtensaMPU { +xtensa_mpu_entry fg[MAX_MPU_FOREGROUND_SEGMENTS ]; +}XtensaMPU ; + typedef struct XtensaGdbReg { int targno; unsigned flags;@@ -526,10 +536,10 @@ struct CPUArchState {uint32_t exclusive_val;
target/xtensa/cpu.h struct CPUArchState { · line 542
Quoted 10 lines
target/xtensa/cpu.h +4
-4#ifndef CONFIG_USER_ONLY -xtensa_tlb_entry itlb[7 ][MAX_TLB_WAY_SIZE ]; -xtensa_tlb_entry dtlb[10 ][MAX_TLB_WAY_SIZE ]; -xtensa_mpu_entry mpu_fg[MAX_MPU_FOREGROUND_SEGMENTS ]; -unsigned autorefill_idx; +union { +XtensaMMU mmu; +XtensaMPU mpu; + };
Is it really worth having this be a union ? I suspect it will make adding migration/savevm support later more awkward.
target/xtensa/cpu.h struct CPUArchState { · line 545
Quoted 3 lines
target/xtensa/cpu.h +0
-0bool runstall;AddressSpace *address_space_er;MemoryRegion *system_er;
Otherwise
thanks
-- PMM
Re: [PATCH 2/2] target/xtensa: tidy TLB way variability logic
In reply to Max FilippovQuoted 62 lines · Max Filippov
Whether TLB ways 5 and 6 are variable is not a property of the TLB instance or a TLB entry instance, it's a property of the xtensa core configuration.
Remove 'varway56' field from thextensa_tlbstructure and remove 'variable' field from thextensa_tlb_entrystructure. Add 'tlb_variable_way' array to the XtensaConfig and use it instead of removed fields.Signed-off-by Max Filippov <[email protected]>--- target/xtensa/cpu.h | 3 +-- target/xtensa/mmu_helper.c | 38 ++++++++++-------------------------- target/xtensa/overlay_tool.h | 15 ++++++++++++-- 3 files changed, 24 insertions(+), 32 deletions(-)target/xtensa/cpu.h +1
-2@@ -316,13 +316,11 @@ typedef struct xtensa_tlb_entry {uint32_t paddr;uint8_t asid;uint8_t attr; -bool variable; } xtensa_tlb_entry;typedef struct xtensa_tlb {unsigned nways;const unsigned way_size[10 ]; -bool varway56;unsigned nrefillentries; }xtensa_tlb ;@@ -493,6 +491,7 @@ typedef struct XtensaConfig {xtensa_tlb itlb;xtensa_tlb dtlb; +bool tlb_variable_way[16 ];uint32_t mpu_align;unsigned n_mpu_fg_segments;target/xtensa/mmu_helper.c +1
-5@@ -105,23 +105,19 @@ static uint32_t xtensa_tlb_get_addr_mask(const CPUXtensaState *env, bool dtlb, uint32_t way) {if (xtensa_option_enabled (env->config,XTENSA_OPTION_MMU )) { -bool varway56 = dtlb ? - env->config->dtlb.varway56 : - env->config->itlb.varway56; -switch (way) {case 4 :return 0xfff00000 <<get_page_size (env, dtlb, way) *2 ;case 5 : -if (varway56) { +if (env->config->tlb_variable_way[5 ]) {return 0xf8000000 <<get_page_size (env, dtlb, way);
target/xtensa/mmu_helper.c static uint32_t xtensa_tlb_get_addr_mask(const CPUXtensaState *env, · line 123
Quoted 10 lines
target/xtensa/mmu_helper.c +1
-1}else {return 0xf8000000 ; }case 6 : -if (varway56) { +if (env->config->tlb_variable_way[6 ]) {return 0xf0000000 << (1 -get_page_size (env, dtlb, way)); }else {return 0xf0000000 ;
So we now have a tlb_variable_way bool for all 16 possible ways, but the code actually only checks it for ways 5 and 6. Should we have an assertion somewhere that the config doesn't try to set it on ways where it has no effect ? Or is there actually a generic behaviour that would make sense for eg "way 3 is variable-way" that we just don't currently implement?
target/xtensa/mmu_helper.c static uint32_t get_vpn_mask(const CPUXtensaState *env, bool dtlb, uint32_t way) · line 153
Quoted 13 lines
target/xtensa/mmu_helper.c +1
-4@@ -150,11 +146,8 @@ static uint32_t get_vpn_mask(const CPUXtensaState *env, bool dtlb, uint32_t way)return xtensa_tlb_get_addr_mask (env, dtlb, way) <<2 ; }else if (way <=6 ) {uint32_t mask =xtensa_tlb_get_addr_mask (env, dtlb, way); -bool varway56 = dtlb ? - env->config->dtlb.varway56 : - env->config->itlb.varway56; -if (varway56) { +if (env->config->tlb_variable_way[5 ]) {return mask << (way ==5 ?2 :3 ); }else {return mask <<1 ;
This doesn't look right -- this branch of the if-else deals with way == 5 and way == 6, but we're only looking at tlb_variable_way[5].
Quoted 21 lines
target/xtensa/mmu_helper.c +1
-5@@ -172,10 +165,6 @@ static void split_tlb_entry_spec_way(const CPUXtensaState *env, uint32_t v, bool dtlb, uint32_t *vpn, uint32_t wi, uint32_t *ei) { - bool varway56 = dtlb ? - env->config->dtlb.varway56 : - env->config->itlb.varway56; - if (!dtlb) { wi &= 7; }@@ -195,7 +184,7 @@ static void split_tlb_entry_spec_way(const CPUXtensaState *env, uint32_t v,break ;case 5 : -if (varway56) { +if (env->config->tlb_variable_way[5 ]) {uint32_t eibase =27 +get_page_size (env, dtlb, wi); *ei = (v >> eibase) &0x3 ; }else {
target/xtensa/mmu_helper.c static void split_tlb_entry_spec_way(const CPUXtensaState *env, uint32_t v, · line 199
Quoted 8 lines
target/xtensa/mmu_helper.c +1
-1@@ -204,7 +193,7 @@ static void split_tlb_entry_spec_way(const CPUXtensaState *env, uint32_t v,break ;case 6 : -if (varway56) { +if (env->config->tlb_variable_way[6 ]) {uint32_t eibase =29 -get_page_size (env, dtlb, wi); *ei = (v >> eibase) &0x7 ; }else {
There's no direct code duplication, but it definitely feels like the logic for "figure out how many bits we're dealing with" is duplicated across these three functions.
I think it ought to be possible to have a function (or maybe two) which take account of both the way number and tlb_get_variable_way[] such that all of these three functions then don't need to have a switch on the way or look at tlb_variable_way[] themselves...
thanks
-- PMM
Re: [PATCH 1/2] target/xtensa: wrap MMU and MPU state into structures
In reply to Peter MaydellQuoted 41 lines · Peter Maydell
Quoted 39 lines · Max Filippov
Make separation of alternative xtensa memory management options state explicit.
Signed-off-by Max Filippov <[email protected]>--- target/xtensa/cpu.h | 18 +++++++++++++---- target/xtensa/mmu_helper.c | 40 +++++++++++++++++++------------------- 2 files changed, 34 insertions(+), 24 deletions(-)target/xtensa/cpu.h +10
-0@@ -326,11 +326,21 @@ typedef struct xtensa_tlb {unsigned nrefillentries; } xtensa_tlb; +typedef struct XtensaMMU { +xtensa_tlb_entry itlb[7 ][MAX_TLB_WAY_SIZE ]; +xtensa_tlb_entry dtlb[10 ][MAX_TLB_WAY_SIZE ]; +unsigned autorefill_idx; +}XtensaMMU ; +typedef struct xtensa_mpu_entry {uint32_t vaddr;uint32_t attr; }xtensa_mpu_entry ; +typedef struct XtensaMPU { +xtensa_mpu_entry fg[MAX_MPU_FOREGROUND_SEGMENTS ]; +}XtensaMPU ; + typedef struct XtensaGdbReg { int targno; unsigned flags;@@ -526,10 +536,10 @@ struct CPUArchState {uint32_t exclusive_val;
Quoted 12 lines
target/xtensa/cpu.h struct CPUArchState { · line 542
Quoted 9 lines
target/xtensa/cpu.h +4
-4#ifndef CONFIG_USER_ONLY -xtensa_tlb_entry itlb[7 ][MAX_TLB_WAY_SIZE ]; -xtensa_tlb_entry dtlb[10 ][MAX_TLB_WAY_SIZE ]; -xtensa_mpu_entry mpu_fg[MAX_MPU_FOREGROUND_SEGMENTS ]; -unsigned autorefill_idx; +union { +XtensaMMU mmu; +XtensaMPU mpu; + };Is it really worth having this be a union ? I suspect it will make adding migration/savevm support later more awkward.
I have a draft implementation of savevm for xtensa and I did this part using subsections with the .needed callback checking whether the MMU or MPU option is enabled in the config. I wonder where the awkwardness is expected.
Quoted 2 lines
Otherwise
Reviewed-by Peter Maydell <[email protected]>
Thanks!
-- Max
Re: [PATCH 2/2] target/xtensa: tidy TLB way variability logic
In reply to Peter MaydellQuoted 68 lines · Peter Maydell
Quoted 66 lines · Max Filippov
Whether TLB ways 5 and 6 are variable is not a property of the TLB instance or a TLB entry instance, it's a property of the xtensa core configuration.
Remove 'varway56' field from thextensa_tlbstructure and remove 'variable' field from thextensa_tlb_entrystructure. Add 'tlb_variable_way' array to the XtensaConfig and use it instead of removed fields.Signed-off-by Max Filippov <[email protected]>--- target/xtensa/cpu.h | 3 +-- target/xtensa/mmu_helper.c | 38 ++++++++++-------------------------- target/xtensa/overlay_tool.h | 15 ++++++++++++-- 3 files changed, 24 insertions(+), 32 deletions(-)target/xtensa/cpu.h +1
-2@@ -316,13 +316,11 @@ typedef struct xtensa_tlb_entry {uint32_t paddr;uint8_t asid;uint8_t attr; -bool variable; } xtensa_tlb_entry;typedef struct xtensa_tlb {unsigned nways;const unsigned way_size[10 ]; -bool varway56;unsigned nrefillentries; }xtensa_tlb ;@@ -493,6 +491,7 @@ typedef struct XtensaConfig {xtensa_tlb itlb;xtensa_tlb dtlb; +bool tlb_variable_way[16 ];uint32_t mpu_align;unsigned n_mpu_fg_segments;target/xtensa/mmu_helper.c +1
-5@@ -105,23 +105,19 @@ static uint32_t xtensa_tlb_get_addr_mask(const CPUXtensaState *env, bool dtlb, uint32_t way) {if (xtensa_option_enabled (env->config,XTENSA_OPTION_MMU )) { -bool varway56 = dtlb ? - env->config->dtlb.varway56 : - env->config->itlb.varway56; -switch (way) {case 4 :return 0xfff00000 <<get_page_size (env, dtlb, way) *2 ;case 5 : -if (varway56) { +if (env->config->tlb_variable_way[5 ]) {return 0xf8000000 <<get_page_size (env, dtlb, way); }else {return 0xf8000000 ; }
Quoted 9 lines
target/xtensa/mmu_helper.c static uint32_t xtensa_tlb_get_addr_mask(const CPUXtensaState *env, · line 123
Quoted 6 lines
target/xtensa/mmu_helper.c +1
-1case 6 : -if (varway56) { +if (env->config->tlb_variable_way[6 ]) {return 0xf0000000 << (1 -get_page_size (env, dtlb, way)); }else {return 0xf0000000 ;So we now have a
tlb_variable_waybool for all 16 possible ways, but the code actually only checks it for ways 5 and 6.
xtensa_tlb_set_entry checks this for all possible ways.
I would say that this is an unfortunate definition of MMU in the xtensa ISA book that uses the variability of the ways 5/6 as a discriminator between MMUv2 and MMUv3.
Quoted 5 lines
Should we have an assertion somewhere that the config doesn't try to set it on ways where it has no effect ? Or is there actually a generic behaviour that would make sense for eg "way 3 is variable-way" that we just don't currently implement?
We currently use the TLB structure to implement the following xtensa memory management options: cacheattr, region protection, region translation, MMUv2 and MMUv3. First three only have one variable way, in MMUv2 all ways except 5 and 6 are variable and in MMUv3 all ways are variable. QEMU supports all of it and tlb_variable_way is set properly in all of these cases.
Quoted 8 lines
Quoted 8 lines
target/xtensa/mmu_helper.c +0
-3@@ -150,11 +146,8 @@ static uint32_t get_vpn_mask(const CPUXtensaState *env, bool dtlb, uint32_t way)return xtensa_tlb_get_addr_mask (env, dtlb, way) <<2 ; }else if (way <=6 ) {uint32_t mask =xtensa_tlb_get_addr_mask (env, dtlb, way); -bool varway56 = dtlb ? - env->config->dtlb.varway56 : - env->config->itlb.varway56;
Quoted 9 lines
target/xtensa/mmu_helper.c static uint32_t get_vpn_mask(const CPUXtensaState *env, bool dtlb, uint32_t way) · line 153
Quoted 5 lines
target/xtensa/mmu_helper.c +1
-1-if (varway56) { +if (env->config->tlb_variable_way[5 ]) {return mask << (way ==5 ?2 :3 ); }else {return mask <<1 ;This doesn't look right -- this branch of the if-else deals with way == 5 and way == 6, but we're only looking at tlb_variable_way[5].
Yeah, that's MMUv2 vs MMUv3 check, again.
Quoted 34 lines
Quoted 21 lines
target/xtensa/mmu_helper.c +1
-5@@ -172,10 +165,6 @@ static void split_tlb_entry_spec_way(const CPUXtensaState *env, uint32_t v, bool dtlb, uint32_t *vpn, uint32_t wi, uint32_t *ei) { - bool varway56 = dtlb ? - env->config->dtlb.varway56 : - env->config->itlb.varway56; - if (!dtlb) { wi &= 7; }@@ -195,7 +184,7 @@ static void split_tlb_entry_spec_way(const CPUXtensaState *env, uint32_t v,break ;case 5 : -if (varway56) { +if (env->config->tlb_variable_way[5 ]) {uint32_t eibase =27 +get_page_size (env, dtlb, wi); *ei = (v >> eibase) &0x3 ; }else {target/xtensa/mmu_helper.c static void split_tlb_entry_spec_way(const CPUXtensaState *env, uint32_t v, · line 199
Quoted 8 lines
target/xtensa/mmu_helper.c +1
-1@@ -204,7 +193,7 @@ static void split_tlb_entry_spec_way(const CPUXtensaState *env, uint32_t v,break ;case 6 : -if (varway56) { +if (env->config->tlb_variable_way[6 ]) {uint32_t eibase =29 -get_page_size (env, dtlb, wi); *ei = (v >> eibase) &0x7 ; }else {There's no direct code duplication, but it definitely feels like the logic for "figure out how many bits we're dealing with" is duplicated across these three functions.
Quoted 4 lines
I think it ought to be possible to have a function (or maybe two) which take account of both the way number and tlb_get_variable_way[] such that all of these three functions then don't need to have a switch on the way or look at tlb_variable_way[] themselves...
Ok, let me have another look at cleaning this up.
Signature · 2 lines
Thanks. -- Max
Re: [PATCH 1/2] target/xtensa: wrap MMU and MPU state into structures
In reply to Max FilippovQuoted 9 lines · Max Filippov
Quoted 7 lines · Peter Maydell
Quoted 4 lines · Max Filippov
target/xtensa/cpu.h +4
-0+union { +XtensaMMU mmu; +XtensaMPU mpu; + };
Quoted 7 lines
Quoted 2 lines
Is it really worth having this be a union ? I suspect it will make adding migration/savevm support later more awkward.
I have a draft implementation of savevm for xtensa and I did this part using subsections with the .needed callback checking whether the MMU or MPU option is enabled in the config. I wonder where the awkwardness is expected.
Oh, well if it works that's fine I guess. I was kind of thinking that if you didn't have the union you could avoid having subsections entirely.
On Arm we don't try to save space in the CPU struct by using unions, even though some fields are A-profile specific and some are R or M-profile specific.
-- PMM