From dc90e95f310f4f821c905b2aec8e9449bb3270fa Mon Sep 17 00:00:00 2001 From: Peter Chubb Date: Wed, 24 Aug 2005 17:13:00 -0700 Subject: [PATCH 1/3] [IA64] Add PAL_VM_SUMMARY/PAL_MEM_ATTRIB to bootloader for SKI This patch implements PAL_VM_SUMMARY (and PAL_MEM_ATTRIB for good measure) and pretends that the simulated machine is a McKinley. Some extra comments and clean-up by Tony Luck. Signed-off-by: Peter Chubb Signed-off-by: Tony Luck --- arch/ia64/hp/sim/boot/boot_head.S | 31 +++++++++++++++++++++++++------ 1 file changed, 25 insertions(+), 6 deletions(-) diff --git a/arch/ia64/hp/sim/boot/boot_head.S b/arch/ia64/hp/sim/boot/boot_head.S index 1c8c7e6a9a5e..a9bd71ac78e2 100644 --- a/arch/ia64/hp/sim/boot/boot_head.S +++ b/arch/ia64/hp/sim/boot/boot_head.S @@ -4,6 +4,7 @@ */ #include +#include .bss .align 16 @@ -49,7 +50,11 @@ GLOBAL_ENTRY(jmp_to_kernel) br.sptk.few b7 END(jmp_to_kernel) - +/* + * r28 contains the index of the PAL function + * r29--31 the args + * Return values in ret0--3 (r8--11) + */ GLOBAL_ENTRY(pal_emulator_static) mov r8=-1 mov r9=256 @@ -62,7 +67,7 @@ GLOBAL_ENTRY(pal_emulator_static) cmp.gtu p6,p7=r9,r28 (p6) br.cond.sptk.few stacked ;; -static: cmp.eq p6,p7=6,r28 /* PAL_PTCE_INFO */ +static: cmp.eq p6,p7=PAL_PTCE_INFO,r28 (p7) br.cond.sptk.few 1f ;; mov r8=0 /* status = 0 */ @@ -70,21 +75,21 @@ static: cmp.eq p6,p7=6,r28 /* PAL_PTCE_INFO */ movl r10=0x0000000200000003 /* count[0], count[1] */ movl r11=0x1000000000002000 /* stride[0], stride[1] */ br.cond.sptk.few rp -1: cmp.eq p6,p7=14,r28 /* PAL_FREQ_RATIOS */ +1: cmp.eq p6,p7=PAL_FREQ_RATIOS,r28 (p7) br.cond.sptk.few 1f mov r8=0 /* status = 0 */ movl r9 =0x100000064 /* proc_ratio (1/100) */ movl r10=0x100000100 /* bus_ratio<<32 (1/256) */ movl r11=0x100000064 /* itc_ratio<<32 (1/100) */ ;; -1: cmp.eq p6,p7=19,r28 /* PAL_RSE_INFO */ +1: cmp.eq p6,p7=PAL_RSE_INFO,r28 (p7) br.cond.sptk.few 1f mov r8=0 /* status = 0 */ mov r9=96 /* num phys stacked */ mov r10=0 /* hints */ mov r11=0 br.cond.sptk.few rp -1: cmp.eq p6,p7=1,r28 /* PAL_CACHE_FLUSH */ +1: cmp.eq p6,p7=PAL_CACHE_FLUSH,r28 /* PAL_CACHE_FLUSH */ (p7) br.cond.sptk.few 1f mov r9=ar.lc movl r8=524288 /* flush 512k million cache lines (16MB) */ @@ -102,7 +107,7 @@ static: cmp.eq p6,p7=6,r28 /* PAL_PTCE_INFO */ mov ar.lc=r9 mov r8=r0 ;; -1: cmp.eq p6,p7=15,r28 /* PAL_PERF_MON_INFO */ +1: cmp.eq p6,p7=PAL_PERF_MON_INFO,r28 (p7) br.cond.sptk.few 1f mov r8=0 /* status = 0 */ movl r9 =0x08122f04 /* generic=4 width=47 retired=8 cycles=18 */ @@ -138,6 +143,20 @@ static: cmp.eq p6,p7=6,r28 /* PAL_PTCE_INFO */ st8 [r29]=r0,16 /* clear remaining bits */ st8 [r18]=r0,16 /* clear remaining bits */ ;; +1: cmp.eq p6,p7=PAL_VM_SUMMARY,r28 +(p7) br.cond.sptk.few 1f + mov r8=0 /* status = 0 */ + movl r9=0x2044040020F1865 /* num_tc_levels=2, num_unique_tcs=4 */ + /* max_itr_entry=64, max_dtr_entry=64 */ + /* hash_tag_id=2, max_pkr=15 */ + /* key_size=24, phys_add_size=50, vw=1 */ + movl r10=0x183C /* rid_size=24, impl_va_msb=60 */ + ;; +1: cmp.eq p6,p7=PAL_MEM_ATTRIB,r28 +(p7) br.cond.sptk.few 1f + mov r8=0 /* status = 0 */ + mov r9=0x80|0x01 /* NatPage|WB */ + ;; 1: br.cond.sptk.few rp stacked: br.ret.sptk.few rp From 714d2dc14914f0f7bb008effe830c99eb47c75df Mon Sep 17 00:00:00 2001 From: Peter Chubb Date: Thu, 25 Aug 2005 17:39:00 -0700 Subject: [PATCH 2/3] [IA64] Allow /proc/pal/cpu0/vm_info under the simulator Not all of the PAL VM calls are implemented for the SKI simulator. Don't just give up if one fails, print information from the calls that succeed. Signed-off-by: Peter Chubb Signed-off-by: Tony Luck --- arch/ia64/kernel/palinfo.c | 127 +++++++++++++++++++------------------ 1 file changed, 66 insertions(+), 61 deletions(-) diff --git a/arch/ia64/kernel/palinfo.c b/arch/ia64/kernel/palinfo.c index 25e7c8344564..89faa603c6be 100644 --- a/arch/ia64/kernel/palinfo.c +++ b/arch/ia64/kernel/palinfo.c @@ -307,11 +307,9 @@ vm_info(char *page) if ((status = ia64_pal_vm_summary(&vm_info_1, &vm_info_2)) !=0) { printk(KERN_ERR "ia64_pal_vm_summary=%ld\n", status); - return 0; - } + } else { - - p += sprintf(p, + p += sprintf(p, "Physical Address Space : %d bits\n" "Virtual Address Space : %d bits\n" "Protection Key Registers(PKR) : %d\n" @@ -319,92 +317,99 @@ vm_info(char *page) "Hash Tag ID : 0x%x\n" "Size of RR.rid : %d\n", vm_info_1.pal_vm_info_1_s.phys_add_size, - vm_info_2.pal_vm_info_2_s.impl_va_msb+1, vm_info_1.pal_vm_info_1_s.max_pkr+1, - vm_info_1.pal_vm_info_1_s.key_size, vm_info_1.pal_vm_info_1_s.hash_tag_id, + vm_info_2.pal_vm_info_2_s.impl_va_msb+1, + vm_info_1.pal_vm_info_1_s.max_pkr+1, + vm_info_1.pal_vm_info_1_s.key_size, + vm_info_1.pal_vm_info_1_s.hash_tag_id, vm_info_2.pal_vm_info_2_s.rid_size); - - if (ia64_pal_mem_attrib(&attrib) != 0) - return 0; - - p += sprintf(p, "Supported memory attributes : "); - sep = ""; - for (i = 0; i < 8; i++) { - if (attrib & (1 << i)) { - p += sprintf(p, "%s%s", sep, mem_attrib[i]); - sep = ", "; - } } - p += sprintf(p, "\n"); + + if (ia64_pal_mem_attrib(&attrib) == 0) { + p += sprintf(p, "Supported memory attributes : "); + sep = ""; + for (i = 0; i < 8; i++) { + if (attrib & (1 << i)) { + p += sprintf(p, "%s%s", sep, mem_attrib[i]); + sep = ", "; + } + } + p += sprintf(p, "\n"); + } if ((status = ia64_pal_vm_page_size(&tr_pages, &vw_pages)) !=0) { printk(KERN_ERR "ia64_pal_vm_page_size=%ld\n", status); - return 0; + } else { + + p += sprintf(p, + "\nTLB walker : %simplemented\n" + "Number of DTR : %d\n" + "Number of ITR : %d\n" + "TLB insertable page sizes : ", + vm_info_1.pal_vm_info_1_s.vw ? "" : "not ", + vm_info_1.pal_vm_info_1_s.max_dtr_entry+1, + vm_info_1.pal_vm_info_1_s.max_itr_entry+1); + + + p = bitvector_process(p, tr_pages); + + p += sprintf(p, "\nTLB purgeable page sizes : "); + + p = bitvector_process(p, vw_pages); } - - p += sprintf(p, - "\nTLB walker : %simplemented\n" - "Number of DTR : %d\n" - "Number of ITR : %d\n" - "TLB insertable page sizes : ", - vm_info_1.pal_vm_info_1_s.vw ? "" : "not ", - vm_info_1.pal_vm_info_1_s.max_dtr_entry+1, - vm_info_1.pal_vm_info_1_s.max_itr_entry+1); - - - p = bitvector_process(p, tr_pages); - - p += sprintf(p, "\nTLB purgeable page sizes : "); - - p = bitvector_process(p, vw_pages); - if ((status=ia64_get_ptce(&ptce)) != 0) { printk(KERN_ERR "ia64_get_ptce=%ld\n", status); - return 0; - } - - p += sprintf(p, + } else { + p += sprintf(p, "\nPurge base address : 0x%016lx\n" "Purge outer loop count : %d\n" "Purge inner loop count : %d\n" "Purge outer loop stride : %d\n" "Purge inner loop stride : %d\n", - ptce.base, ptce.count[0], ptce.count[1], ptce.stride[0], ptce.stride[1]); + ptce.base, ptce.count[0], ptce.count[1], + ptce.stride[0], ptce.stride[1]); - p += sprintf(p, + p += sprintf(p, "TC Levels : %d\n" "Unique TC(s) : %d\n", vm_info_1.pal_vm_info_1_s.num_tc_levels, vm_info_1.pal_vm_info_1_s.max_unique_tcs); - for(i=0; i < vm_info_1.pal_vm_info_1_s.num_tc_levels; i++) { - for (j=2; j>0 ; j--) { - tc_pages = 0; /* just in case */ + for(i=0; i < vm_info_1.pal_vm_info_1_s.num_tc_levels; i++) { + for (j=2; j>0 ; j--) { + tc_pages = 0; /* just in case */ - /* even without unification, some levels may not be present */ - if ((status=ia64_pal_vm_info(i,j, &tc_info, &tc_pages)) != 0) { - continue; - } + /* even without unification, some levels may not be present */ + if ((status=ia64_pal_vm_info(i,j, &tc_info, &tc_pages)) != 0) { + continue; + } - p += sprintf(p, + p += sprintf(p, "\n%s Translation Cache Level %d:\n" "\tHash sets : %d\n" "\tAssociativity : %d\n" "\tNumber of entries : %d\n" "\tFlags : ", - cache_types[j+tc_info.tc_unified], i+1, tc_info.tc_num_sets, - tc_info.tc_associativity, tc_info.tc_num_entries); + cache_types[j+tc_info.tc_unified], i+1, + tc_info.tc_num_sets, + tc_info.tc_associativity, + tc_info.tc_num_entries); - if (tc_info.tc_pf) p += sprintf(p, "PreferredPageSizeOptimized "); - if (tc_info.tc_unified) p += sprintf(p, "Unified "); - if (tc_info.tc_reduce_tr) p += sprintf(p, "TCReduction"); + if (tc_info.tc_pf) + p += sprintf(p, "PreferredPageSizeOptimized "); + if (tc_info.tc_unified) + p += sprintf(p, "Unified "); + if (tc_info.tc_reduce_tr) + p += sprintf(p, "TCReduction"); - p += sprintf(p, "\n\tSupported page sizes: "); + p += sprintf(p, "\n\tSupported page sizes: "); - p = bitvector_process(p, tc_pages); + p = bitvector_process(p, tc_pages); - /* when unified date (j=2) is enough */ - if (tc_info.tc_unified) break; + /* when unified date (j=2) is enough */ + if (tc_info.tc_unified) + break; + } } } p += sprintf(p, "\n"); @@ -440,14 +445,14 @@ register_info(char *page) p += sprintf(p, "\n"); } - if (ia64_pal_rse_info(&phys_stacked, &hints) != 0) return 0; + if (ia64_pal_rse_info(&phys_stacked, &hints) == 0) { p += sprintf(p, "RSE stacked physical registers : %ld\n" "RSE load/store hints : %ld (%s)\n", phys_stacked, hints.ph_data, hints.ph_data < RSE_HINTS_COUNT ? rse_hints[hints.ph_data]: "(??)"); - + } if (ia64_pal_debug_info(&iregs, &dregs)) return 0; From 6cf07a8cc86a0b471466c7fe45892f7ef434015b Mon Sep 17 00:00:00 2001 From: Peter Chubb Date: Tue, 23 Aug 2005 20:07:00 -0700 Subject: [PATCH 3/3] [IA64] Fix nasty VMLPT problem... I've solved the problem I was having with the simulator and not booting Debian. The problem is that the number of bits for the virtual linear array short-format VHPT (Virtually mapped linear page table, VMLPT for short) is being tested incorrectly. There are two problems: 1. The PAL call that should tell the kernel the size of the virtual address space isn't implemented for the simulator, so the kernel uses the default 50. This is addressed separately in dc90e95f310f4f821c905b2aec8e9449bb3270fa 2. In arch/ia64/mm/init.c there's code to calcualte the size of the VMLPT based on the number of implemented virtual address bits and the page size. It checks to see if the VMLPT base address overlaps the top of the mapped region, but this check doesn't allow for the address space hole, and in fact will never trigger. Here's an alternative test and panic, that I think is more accurate. Signed-off-by: Peter Chubb Signed-off-by: Tony Luck --- arch/ia64/mm/init.c | 15 ++++++++++++--- 1 file changed, 12 insertions(+), 3 deletions(-) diff --git a/arch/ia64/mm/init.c b/arch/ia64/mm/init.c index 65f9958db9f0..1281c609ee98 100644 --- a/arch/ia64/mm/init.c +++ b/arch/ia64/mm/init.c @@ -382,13 +382,22 @@ ia64_mmu_init (void *my_cpu_data) if (impl_va_bits < 51 || impl_va_bits > 61) panic("CPU has bogus IMPL_VA_MSB value of %lu!\n", impl_va_bits - 1); + /* + * mapped_space_bits - PAGE_SHIFT is the total number of ptes we need, + * which must fit into "vmlpt_bits - pte_bits" slots. Second half of + * the test makes sure that our mapped space doesn't overlap the + * unimplemented hole in the middle of the region. + */ + if ((mapped_space_bits - PAGE_SHIFT > vmlpt_bits - pte_bits) || + (mapped_space_bits > impl_va_bits - 1)) + panic("Cannot build a big enough virtual-linear page table" + " to cover mapped address space.\n" + " Try using a smaller page size.\n"); + /* place the VMLPT at the end of each page-table mapped region: */ pta = POW2(61) - POW2(vmlpt_bits); - if (POW2(mapped_space_bits) >= pta) - panic("mm/init: overlap between virtually mapped linear page table and " - "mapped kernel space!"); /* * Set the (virtually mapped linear) page table address. Bit * 8 selects between the short and long format, bits 2-7 the