diff options
38 files changed, 843 insertions, 865 deletions
diff --git a/arch/ppc/kernel/cputable.c b/arch/ppc/kernel/cputable.c index 546e1ea4caf..6b76cf58d9e 100644 --- a/arch/ppc/kernel/cputable.c +++ b/arch/ppc/kernel/cputable.c @@ -91,7 +91,7 @@ struct cpu_spec cpu_specs[] = { .cpu_features = CPU_FTR_COMMON | CPU_FTR_601 | CPU_FTR_HPTE_TABLE, .cpu_user_features = COMMON_PPC | PPC_FEATURE_601_INSTR | - PPC_FEATURE_UNIFIED_CACHE, + PPC_FEATURE_UNIFIED_CACHE | PPC_FEATURE_NO_TB, .icache_bsize = 32, .dcache_bsize = 32, .cpu_setup = __setup_cpu_601 @@ -745,7 +745,8 @@ struct cpu_spec cpu_specs[] = { .cpu_name = "403GCX", .cpu_features = CPU_FTR_SPLIT_ID_CACHE | CPU_FTR_USE_TB, - .cpu_user_features = PPC_FEATURE_32 | PPC_FEATURE_HAS_MMU, + .cpu_user_features = PPC_FEATURE_32 | + PPC_FEATURE_HAS_MMU | PPC_FEATURE_NO_TB, .icache_bsize = 16, .dcache_bsize = 16, }, diff --git a/arch/sparc64/kernel/dtlb_base.S b/arch/sparc64/kernel/dtlb_base.S index 702d349c1e8..6528786840c 100644 --- a/arch/sparc64/kernel/dtlb_base.S +++ b/arch/sparc64/kernel/dtlb_base.S @@ -53,19 +53,18 @@ * be guaranteed to be 0 ... mmu_context.h does guarantee this * by only using 10 bits in the hwcontext value. */ -#define CREATE_VPTE_OFFSET1(r1, r2) +#define CREATE_VPTE_OFFSET1(r1, r2) nop #define CREATE_VPTE_OFFSET2(r1, r2) \ srax r1, 10, r2 -#define CREATE_VPTE_NOP nop #else #define CREATE_VPTE_OFFSET1(r1, r2) \ srax r1, PAGE_SHIFT, r2 #define CREATE_VPTE_OFFSET2(r1, r2) \ sllx r2, 3, r2 -#define CREATE_VPTE_NOP #endif /* DTLB ** ICACHE line 1: Quick user TLB misses */ + mov TLB_SFSR, %g1 ldxa [%g1 + %g1] ASI_DMMU, %g4 ! Get TAG_ACCESS andcc %g4, TAG_CONTEXT_BITS, %g0 ! From Nucleus? from_tl1_trap: @@ -74,18 +73,16 @@ from_tl1_trap: be,pn %xcc, kvmap ! Yep, special processing CREATE_VPTE_OFFSET2(%g4, %g6) ! Create VPTE offset cmp %g5, 4 ! Last trap level? - be,pn %xcc, longpath ! Yep, cannot risk VPTE miss - nop ! delay slot /* DTLB ** ICACHE line 2: User finish + quick kernel TLB misses */ + be,pn %xcc, longpath ! Yep, cannot risk VPTE miss + nop ! delay slot ldxa [%g3 + %g6] ASI_S, %g5 ! Load VPTE 1: brgez,pn %g5, longpath ! Invalid, branch out nop ! Delay-slot 9: stxa %g5, [%g0] ASI_DTLB_DATA_IN ! Reload TLB retry ! Trap return nop - nop - nop /* DTLB ** ICACHE line 3: winfixups+real_faults */ longpath: @@ -106,8 +103,7 @@ longpath: nop nop nop - CREATE_VPTE_NOP + nop #undef CREATE_VPTE_OFFSET1 #undef CREATE_VPTE_OFFSET2 -#undef CREATE_VPTE_NOP diff --git a/arch/sparc64/kernel/dtlb_prot.S b/arch/sparc64/kernel/dtlb_prot.S index d848bb7374b..e0a92016260 100644 --- a/arch/sparc64/kernel/dtlb_prot.S +++ b/arch/sparc64/kernel/dtlb_prot.S @@ -14,14 +14,14 @@ */ /* PROT ** ICACHE line 1: User DTLB protection trap */ - stxa %g0, [%g1] ASI_DMMU ! Clear SFSR FaultValid bit - membar #Sync ! Synchronize ASI stores - rdpr %pstate, %g5 ! Move into alternate globals + mov TLB_SFSR, %g1 + stxa %g0, [%g1] ASI_DMMU ! Clear FaultValid bit + membar #Sync ! Synchronize stores + rdpr %pstate, %g5 ! Move into alt-globals wrpr %g5, PSTATE_AG|PSTATE_MG, %pstate - rdpr %tl, %g1 ! Need to do a winfixup? + rdpr %tl, %g1 ! Need a winfixup? cmp %g1, 1 ! Trap level >1? - mov TLB_TAG_ACCESS, %g4 ! Prepare reload of vaddr - nop + mov TLB_TAG_ACCESS, %g4 ! For reload of vaddr /* PROT ** ICACHE line 2: More real fault processing */ bgu,pn %xcc, winfix_trampoline ! Yes, perform winfixup diff --git a/arch/sparc64/kernel/head.S b/arch/sparc64/kernel/head.S index 4c942f71184..b49dcd4504b 100644 --- a/arch/sparc64/kernel/head.S +++ b/arch/sparc64/kernel/head.S @@ -28,19 +28,14 @@ #include <asm/mmu.h> /* This section from from _start to sparc64_boot_end should fit into - * 0x0000.0000.0040.4000 to 0x0000.0000.0040.8000 and will be sharing space - * with bootup_user_stack, which is from 0x0000.0000.0040.4000 to - * 0x0000.0000.0040.6000 and empty_bad_page, which is from - * 0x0000.0000.0040.6000 to 0x0000.0000.0040.8000. + * 0x0000000000404000 to 0x0000000000408000. */ - .text .globl start, _start, stext, _stext _start: start: _stext: stext: -bootup_user_stack: ! 0x0000000000404000 b sparc64_boot flushw /* Flush register file. */ @@ -392,31 +387,30 @@ tlb_fixup_done: * former does use this code, the latter does not yet due * to some complexities. That should be fixed up at some * point. + * + * There used to be enormous complexity wrt. transferring + * over from the firwmare's trap table to the Linux kernel's. + * For example, there was a chicken & egg problem wrt. building + * the OBP page tables, yet needing to be on the Linux kernel + * trap table (to translate PAGE_OFFSET addresses) in order to + * do that. + * + * We now handle OBP tlb misses differently, via linear lookups + * into the prom_trans[] array. So that specific problem no + * longer exists. Yet, unfortunately there are still some issues + * preventing trampoline.S from using this code... ho hum. */ .globl setup_trap_table setup_trap_table: save %sp, -192, %sp - /* Force interrupts to be disabled. Transferring over to - * the Linux trap table is a very delicate operation. - * Until we are actually on the Linux trap table, we cannot - * get the PAGE_OFFSET linear mappings translated. We need - * that mapping to be setup in order to initialize the firmware - * page tables. - * - * So there is this window of time, from the return from - * prom_set_trap_table() until inherit_prom_mappings_post() - * (in arch/sparc64/mm/init.c) completes, during which no - * firmware address space accesses can be made. - */ + /* Force interrupts to be disabled. */ rdpr %pstate, %o1 andn %o1, PSTATE_IE, %o1 wrpr %o1, 0x0, %pstate wrpr %g0, 15, %pil - /* Ok, now make the final valid firmware call to jump over - * to the Linux trap table. - */ + /* Make the firmware call to jump over to the Linux trap table. */ call prom_set_trap_table sethi %hi(sparc64_ttable_tl0), %o0 @@ -540,15 +534,21 @@ setup_tba: /* i0 = is_starfire */ ret restore +sparc64_boot_end: + +#include "systbls.S" +#include "ktlb.S" +#include "etrap.S" +#include "rtrap.S" +#include "winfixup.S" +#include "entry.S" /* - * The following skips make sure the trap table in ttable.S is aligned + * The following skip makes sure the trap table in ttable.S is aligned * on a 32K boundary as required by the v9 specs for TBA register. */ -sparc64_boot_end: - .skip 0x2000 + _start - sparc64_boot_end -bootup_user_stack_end: - .skip 0x2000 +1: + .skip 0x4000 + _start - 1b #ifdef CONFIG_SBUS /* This is just a hack to fool make depend config.h discovering @@ -560,15 +560,6 @@ bootup_user_stack_end: ! 0x0000000000408000 #include "ttable.S" -#include "systbls.S" -#include "ktlb.S" -#include "etrap.S" -#include "rtrap.S" -#include "winfixup.S" -#include "entry.S" - - /* This is just anal retentiveness on my part... */ - .align 16384 .data .align 8 diff --git a/arch/sparc64/kernel/itlb_base.S b/arch/sparc64/kernel/itlb_base.S index b5e32dfa4fb..4951ff8f687 100644 --- a/arch/sparc64/kernel/itlb_base.S +++ b/arch/sparc64/kernel/itlb_base.S @@ -15,14 +15,12 @@ */ #define CREATE_VPTE_OFFSET1(r1, r2) \ srax r1, 10, r2 -#define CREATE_VPTE_OFFSET2(r1, r2) -#define CREATE_VPTE_NOP nop +#define CREATE_VPTE_OFFSET2(r1, r2) nop #else /* PAGE_SHIFT */ #define CREATE_VPTE_OFFSET1(r1, r2) \ srax r1, PAGE_SHIFT, r2 #define CREATE_VPTE_OFFSET2(r1, r2) \ sllx r2, 3, r2 -#define CREATE_VPTE_NOP #endif /* PAGE_SHIFT */ @@ -36,6 +34,7 @@ */ /* ITLB ** ICACHE line 1: Quick user TLB misses */ + mov TLB_SFSR, %g1 ldxa [%g1 + %g1] ASI_IMMU, %g4 ! Get TAG_ACCESS CREATE_VPTE_OFFSET1(%g4, %g6) ! Create VPTE offset CREATE_VPTE_OFFSET2(%g4, %g6) ! Create VPTE offset @@ -43,41 +42,38 @@ 1: brgez,pn %g5, 3f ! Not valid, branch out sethi %hi(_PAGE_EXEC), %g4 ! Delay-slot andcc %g5, %g4, %g0 ! Executable? + +/* ITLB ** ICACHE line 2: Real faults */ be,pn %xcc, 3f ! Nope, branch. nop ! Delay-slot 2: stxa %g5, [%g0] ASI_ITLB_DATA_IN ! Load PTE into TLB retry ! Trap return -3: rdpr %pstate, %g4 ! Move into alternate globals - -/* ITLB ** ICACHE line 2: Real faults */ +3: rdpr %pstate, %g4 ! Move into alt-globals wrpr %g4, PSTATE_AG|PSTATE_MG, %pstate rdpr %tpc, %g5 ! And load faulting VA mov FAULT_CODE_ITLB, %g4 ! It was read from ITLB -sparc64_realfault_common: ! Called by TL0 dtlb_miss too + +/* ITLB ** ICACHE line 3: Finish faults */ +sparc64_realfault_common: ! Called by dtlb_miss stb %g4, [%g6 + TI_FAULT_CODE] stx %g5, [%g6 + TI_FAULT_ADDR] ba,pt %xcc, etrap ! Save state 1: rd %pc, %g7 ! ... - nop - -/* ITLB ** ICACHE line 3: Finish faults + window fixups */ call do_sparc64_fault ! Call fault handler add %sp, PTREGS_OFF, %o0! Compute pt_regs arg ba,pt %xcc, rtrap_clr_l6 ! Restore cpu state nop + +/* ITLB ** ICACHE line 4: Window fixups */ winfix_trampoline: rdpr %tpc, %g3 ! Prepare winfixup TNPC - or %g3, 0x7c, %g3 ! Compute offset to branch + or %g3, 0x7c, %g3 ! Compute branch offset wrpr %g3, %tnpc ! Write it into TNPC done ! Do it to it - -/* ITLB ** ICACHE line 4: Unused... */ nop nop nop nop - CREATE_VPTE_NOP #undef CREATE_VPTE_OFFSET1 #undef CREATE_VPTE_OFFSET2 -#undef CREATE_VPTE_NOP diff --git a/arch/sparc64/kernel/ktlb.S b/arch/sparc64/kernel/ktlb.S index 7796b37f478..d9244d3c9f7 100644 --- a/arch/sparc64/kernel/ktlb.S +++ b/arch/sparc64/kernel/ktlb.S @@ -58,9 +58,6 @@ vpte_noent: done vpte_insn_obp: - sethi %hi(prom_pmd_phys), %g5 - ldx [%g5 + %lo(prom_pmd_phys)], %g5 - /* Behave as if we are at TL0. */ wrpr %g0, 1, %tl rdpr %tpc, %g4 /* Find original faulting iaddr */ @@ -71,58 +68,57 @@ vpte_insn_obp: mov TLB_SFSR, %g1 stxa %g4, [%g1 + %g1] ASI_IMMU - /* Get PMD offset. */ - srlx %g4, 23, %g6 - and %g6, 0x7ff, %g6 - sllx %g6, 2, %g6 - - /* Load PMD, is it valid? */ - lduwa [%g5 + %g6] ASI_PHYS_USE_EC, %g5 - brz,pn %g5, longpath - sllx %g5, 11, %g5 - - /* Get PTE offset. */ - srlx %g4, 13, %g6 - and %g6, 0x3ff, %g6 - sllx %g6, 3, %g6 - - /* Load PTE. */ - ldxa [%g5 + %g6] ASI_PHYS_USE_EC, %g5 - brgez,pn %g5, longpath - nop - - /* TLB load and return from trap. */ + sethi %hi(prom_trans), %g5 + or %g5, %lo(prom_trans), %g5 + +1: ldx [%g5 + 0x00], %g6 ! base + brz,a,pn %g6, longpath ! no more entries, fail + mov TLB_SFSR, %g1 ! and restore %g1 + ldx [%g5 + 0x08], %g1 ! len + add %g6, %g1, %g1 ! end + cmp %g6, %g4 + bgu,pt %xcc, 2f + cmp %g4, %g1 + bgeu,pt %xcc, 2f + ldx [%g5 + 0x10], %g1 ! PTE + + /* TLB load, restore %g1, and return from trap. */ + sub %g4, %g6, %g6 + add %g1, %g6, %g5 + mov TLB_SFSR, %g1 stxa %g5, [%g0] ASI_ITLB_DATA_IN retry -kvmap_do_obp: - sethi %hi(prom_pmd_phys), %g5 - ldx [%g5 + %lo(prom_pmd_phys)], %g5 - - /* Get PMD offset. */ - srlx %g4, 23, %g6 - and %g6, 0x7ff, %g6 - sllx %g6, 2, %g6 - - /* Load PMD, is it valid? */ - lduwa [%g5 + %g6] ASI_PHYS_USE_EC, %g5 - brz,pn %g5, longpath - sllx %g5, 11, %g5 - - /* Get PTE offset. */ - srlx %g4, 13, %g6 - and %g6, 0x3ff, %g6 - sllx %g6, 3, %g6 - - /* Load PTE. */ - ldxa [%g5 + %g6] ASI_PHYS_USE_EC, %g5 - brgez,pn %g5, longpath - nop +2: ba,pt %xcc, 1b + add %g5, (3 * 8), %g5 ! next entry - /* TLB load and return from trap. */ +kvmap_do_obp: + sethi %hi(prom_trans), %g5 + or %g5, %lo(prom_trans), %g5 + srlx %g4, 13, %g4 + sllx %g4, 13, %g4 + +1: ldx [%g5 + 0x00], %g6 ! base + brz,a,pn %g6, longpath ! no more entries, fail + mov TLB_SFSR, %g1 ! and restore %g1 + ldx [%g5 + 0x08], %g1 ! len + add %g6, %g1, %g1 ! end + cmp %g6, %g4 + bgu,pt %xcc, 2f + cmp %g4, %g1 + bgeu,pt %xcc, 2f + ldx [%g5 + 0x10], %g1 ! PTE + + /* TLB load, restore %g1, and return from trap. */ + sub %g4, %g6, %g6 + add %g1, %g6, %g5 + mov TLB_SFSR, %g1 stxa %g5, [%g0] ASI_DTLB_DATA_IN retry +2: ba,pt %xcc, 1b + add %g5, (3 * 8), %g5 ! next entry + /* * On a first level data miss, check whether this is to the OBP range (note * that such accesses can be made by prom, as well as by kernel using diff --git a/arch/sparc64/mm/init.c b/arch/sparc64/mm/init.c index 0d2e967c720..1e44ee26cee 100644 --- a/arch/sparc64/mm/init.c +++ b/arch/sparc64/mm/init.c @@ -105,7 +105,7 @@ static void __init read_obp_memory(const char *property, regs[i].phys_addr = base; regs[i].reg_size = size; } - sort(regs, ents, sizeof(struct linux_prom64_registers), + sort(regs, ents, sizeof(struct linux_prom64_registers), cmp_p64, NULL); } @@ -367,8 +367,11 @@ struct linux_prom_translation { unsigned long size; unsigned long data; }; -static struct linux_prom_translation prom_trans[512] __initdata; -static unsigned int prom_trans_ents __initdata; + +/* Exported for kernel TLB miss handling in ktlb.S */ +struct linux_prom_translation prom_trans[512] __read_mostly; +unsigned int prom_trans_ents __read_mostly; +unsigned int swapper_pgd_zero __read_mostly; extern unsigned long prom_boot_page; extern void prom_remap(unsigned long physpage, unsigned long virtpage, int mmu_ihandle); @@ -378,122 +381,57 @@ extern void register_prom_callbacks(void); /* Exported for SMP bootup purposes. */ unsigned long kern_locked_tte_data; -/* Exported for kernel TLB miss handling in ktlb.S */ -unsigned long prom_pmd_phys __read_mostly; -unsigned int swapper_pgd_zero __read_mostly; - -static pmd_t *prompmd __read_mostly; - -#define BASE_PAGE_SIZE 8192 - /* * Translate PROM's mapping we capture at boot time into physical address. * The second parameter is only set from prom_callback() invocations. */ unsigned long prom_virt_to_phys(unsigned long promva, int *error) { - pmd_t *pmdp = prompmd + ((promva >> 23) & 0x7ff); - pte_t *ptep; - unsigned long base; - - if (pmd_none(*pmdp)) { - if (error) - *error = 1; - return 0; - } - ptep = (pte_t *)__pmd_page(*pmdp) + ((promva >> 13) & 0x3ff); - if (!pte_present(*ptep)) { - if (error) - *error = 1; - return 0; - } - if (error) { - *error = 0; - return pte_val(*ptep); - } - base = pte_val(*ptep) & _PAGE_PADDR; - - return base + (promva & (BASE_PAGE_SIZE - 1)); -} + int i; -/* The obp translations are saved based on 8k pagesize, since obp can - * use a mixture of pagesizes. Misses to the LOW_OBP_ADDRESS -> - * HI_OBP_ADDRESS range are handled in entry.S and do not use the vpte - * scheme (also, see rant in inherit_locked_prom_mappings()). - */ -static void __init build_obp_range(unsigned long start, unsigned long end, unsigned long data) -{ - unsigned long vaddr; + for (i = 0; i < prom_trans_ents; i++) { + struct linux_prom_translation *p = &prom_trans[i]; - for (vaddr = start; vaddr < end; vaddr += BASE_PAGE_SIZE) { - unsigned long val; - pmd_t *pmd; - pte_t *pte; + if (promva >= p->virt && + promva < (p->virt + p->size)) { + unsigned long base = p->data & _PAGE_PADDR; - pmd = prompmd + ((vaddr >> 23) & 0x7ff); - if (pmd_none(*pmd)) { - pte = __alloc_bootmem(BASE_PAGE_SIZE, BASE_PAGE_SIZE, - PAGE_SIZE); - if (!pte) - prom_halt(); - memset(pte, 0, BASE_PAGE_SIZE); - pmd_set(pmd, pte); + if (error) + *error = 0; + return base + (promva & (8192 - 1)); } - pte = (pte_t *) __pmd_page(*pmd) + ((vaddr >> 13) & 0x3ff); - - val = data; - - /* Clear diag TTE bits. */ - if (tlb_type == spitfire) - val &= ~0x0003fe0000000000UL; - - set_pte_at(&init_mm, vaddr, pte, - __pte(val | _PAGE_MODIFIED)); - - data += BASE_PAGE_SIZE; } + if (error) + *error = 1; + return 0UL; } +/* The obp translations are saved based on 8k pagesize, since obp can + * use a mixture of pagesizes. Misses to the LOW_OBP_ADDRESS -> + * HI_OBP_ADDRESS range are handled in ktlb.S and do not use the vpte + * scheme (also, see rant in inherit_locked_prom_mappings()). + */ static inline int in_obp_range(unsigned long vaddr) { return (vaddr >= LOW_OBP_ADDRESS && vaddr < HI_OBP_ADDRESS); } -#define OBP_PMD_SIZE 2048 -static void __init build_obp_pgtable(void) +static int cmp_ptrans(const void *a, const void *b) { - unsigned long i; - - prompmd = __alloc_bootmem(OBP_PMD_SIZE, OBP_PMD_SIZE, PAGE_SIZE); - if (!prompmd) - prom_halt(); - - memset(prompmd, 0, OBP_PMD_SIZE); - - prom_pmd_phys = __pa(prompmd); - - for (i = 0; i < prom_trans_ents; i++) { - unsigned long start, end; - - if (!in_obp_range(prom_trans[i].virt)) - continue; + const struct linux_prom_translation *x = a, *y = b; - start = prom_trans[i].virt; - end = start + prom_trans[i].size; - if (end > HI_OBP_ADDRESS) - end = HI_OBP_ADDRESS; - - build_obp_range(start, end, prom_trans[i].data); - } + if (x->virt > y->virt) + return 1; + if (x->virt < y->virt) + return -1; + return 0; } -/* Read OBP translations property into 'prom_trans[]'. - * Return the number of entries. - */ +/* Read OBP translations property into 'prom_trans[]'. */ static void __init read_obp_translations(void) { - int n, node; + int n, node, ents, first, last, i; node = prom_finddevice("/virtual-memory"); n = prom_getproplen(node, "translations"); @@ -515,7 +453,41 @@ static void __init read_obp_translations(void) n = n / sizeof(struct linux_prom_translation); - prom_trans_ents = n; + ents = n; + + sort(prom_trans, ents, sizeof(struct linux_prom_translation), + cmp_ptrans, NULL); + + /* Now kick out all the non-OBP entries. */ + for (i = 0; i < ents; i++) { + if (in_obp_range(prom_trans[i].virt)) + break; + } + first = i; + for (; i < ents; i++) { + if (!in_obp_range(prom_trans[i].virt)) + break; + } + last = i; + + for (i = 0; i < (last - first); i++) { + struct linux_prom_translation *src = &prom_trans[i + first]; + struct linux_prom_translation *dest = &prom_trans[i]; + + *dest = *src; + } + for (; i < ents; i++) { + struct linux_prom_translation *dest = &prom_trans[i]; + dest->virt = dest->size = dest->data = 0x0UL; + } + + prom_trans_ents = last - first; + + if (tlb_type == spitfire) { + /* Clear diag TTE bits. */ + for (i = 0; i < prom_trans_ents; i++) + prom_trans[i].data &= ~0x0003fe0000000000UL; + } } static void __init remap_kernel(void) @@ -553,21 +525,18 @@ static void __init remap_kernel(void) } -static void __init inherit_prom_mappings_pre(void) +static void __init inherit_prom_mappings(void) { read_obp_translations(); /* Now fixup OBP's idea about where we really are mapped. */ prom_printf("Remapping the kernel... "); remap_kernel(); - prom_printf("done.\n"); -} -static void __init inherit_prom_mappings_post(void) -{ - build_obp_pgtable(); + prom_printf("Registering callbacks... "); register_prom_callbacks(); + prom_printf("done.\n"); } /* The OBP specifications for sun4u mark 0xfffffffc00000000 and @@ -1519,7 +1488,7 @@ void __init paging_init(void) swapper_pgd_zero = pgd_val(swapper_pg_dir[0]); - inherit_prom_mappings_pre(); + inherit_prom_mappings(); /* Ok, we can use our TLB miss and window trap handlers safely. * We need to do a quick peek here to see if we are on StarFire @@ -1530,23 +1499,15 @@ void __init paging_init(void) extern void setup_tba(int); setup_tba(this_is_starfire); } - __flush_tlb_all(); - /* Everything from this point forward, until we are done with - * inherit_prom_mappings_post(), must complete successfully - * without calling into the firmware. The firwmare page tables - * have not been built, but we are running on the Linux kernel's - * trap table. - */ + inherit_locked_prom_mappings(1); + + __flush_tlb_all(); /* Setup bootmem... */ pages_avail = 0; last_valid_pfn = end_pfn = bootmem_init(&pages_avail); - inherit_prom_mappings_post(); - - inherit_locked_prom_mappings(1); - #ifdef CONFIG_DEBUG_PAGEALLOC kernel_physical_mapping_init(); #endif diff --git a/arch/um/drivers/Makefile b/arch/um/drivers/Makefile index 783e18cae09..de17d4c6e02 100644 --- a/arch/um/drivers/Makefile +++ b/arch/um/drivers/Makefile @@ -13,7 +13,7 @@ mcast-objs := mcast_kern.o mcast_user.o net-objs := net_kern.o net_user.o mconsole-objs := mconsole_kern.o mconsole_user.o hostaudio-objs := hostaudio_kern.o -ubd-objs := ubd_kern.o +ubd-objs := ubd_kern.o ubd_user.o port-objs := port_kern.o port_user.o harddog-objs := harddog_kern.o harddog_user.o diff --git a/arch/um/drivers/ubd_kern.c b/arch/um/drivers/ubd_kern.c index e77a38da435..f73134333f6 100644 --- a/arch/um/drivers/ubd_kern.c +++ b/arch/um/drivers/ubd_kern.c @@ -35,7 +35,6 @@ #include "linux/blkpg.h" #include "linux/genhd.h" #include "linux/spinlock.h" -#include "asm/atomic.h" #include "asm/segment.h" #include "asm/uaccess.h" #include "asm/irq.h" @@ -54,21 +53,20 @@ #include "mem.h" #include "mem_kern.h" #include "cow.h" -#include "aio.h" enum ubd_req { UBD_READ, UBD_WRITE }; struct io_thread_req { - enum aio_type op; + enum ubd_req op; int fds[2]; unsigned long offsets[2]; unsigned long long offset; unsigned long length; char *buffer; int sectorsize; - int bitmap_offset; - long bitmap_start; - long bitmap_end; + unsigned long sector_mask; + unsigned long long cow_offset; + unsigned long bitmap_words[2]; int error; }; @@ -82,31 +80,28 @@ extern int create_cow_file(char *cow_file, char *backing_file, unsigned long *bitmap_len_out, int *data_offset_out); extern int read_cow_bitmap(int fd, void *buf, int offset, int len); -extern void do_io(struct io_thread_req *req, struct request *r, - unsigned long *bitmap); +extern void do_io(struct io_thread_req *req); -static inline int ubd_test_bit(__u64 bit, void *data) +static inline int ubd_test_bit(__u64 bit, unsigned char *data) { - unsigned char *buffer = data; __u64 n; int bits, off; - bits = sizeof(buffer[0]) * 8; + bits = sizeof(data[0]) * 8; n = bit / bits; off = bit % bits; - return((buffer[n] & (1 << off)) != 0); + return((data[n] & (1 << off)) != 0); } -static inline void ubd_set_bit(__u64 bit, void *data) +static inline void ubd_set_bit(__u64 bit, unsigned char *data) { - unsigned char *buffer = data; __u64 n; int bits, off; - bits = sizeof(buffer[0]) * 8; + bits = sizeof(data[0]) * 8; n = bit / bits; off = bit % bits; - buffer[n] |= (1 << off); + data[n] |= (1 << off); } /*End stuff from ubd_user.h*/ @@ -115,6 +110,8 @@ static inline void ubd_set_bit(__u64 bit, void *data) static DEFINE_SPINLOCK(ubd_io_lock); static DEFINE_SPINLOCK(ubd_lock); +static void (*do_ubd)(void); + static int ubd_open(struct inode * inode, struct file * filp); static int ubd_release(struct inode * inode, struct file * file); static int ubd_ioctl(struct inode * inode, struct file * file, @@ -161,8 +158,6 @@ struct cow { int data_offset; }; -#define MAX_SG 64 - struct ubd { char *file; int count; @@ -173,7 +168,6 @@ struct ubd { int no_cow; struct cow cow; struct platform_device pdev; - struct scatterlist sg[MAX_SG]; }; #define DEFAULT_COW { \ @@ -466,114 +460,81 @@ __uml_help(fakehd, ); static void do_ubd_request(request_queue_t * q); -static int in_ubd; + +/* Only changed by ubd_init, which is an initcall. */ +int thread_fd = -1; /* Changed by ubd_handler, which is serialized because interrupts only * happen on CPU 0. */ int intr_count = 0; -static void ubd_end_request(struct request *req, int bytes, int uptodate) +/* call ubd_finish if you need to serialize */ +static void __ubd_finish(struct request *req, int error) { - if (!end_that_request_first(req, uptodate, bytes >> 9)) { - add_disk_randomness(req->rq_disk); - end_that_request_last(req); + int nsect; + + if(error){ + end_request(req, 0); + return; } + nsect = req->current_nr_sectors; + req->sector += nsect; + req->buffer += nsect << 9; + req->errors = 0; + req->nr_sectors -= nsect; + req->current_nr_sectors = 0; + end_request(req, 1); } -/* call ubd_finish if you need to serialize */ -static void __ubd_finish(struct request *req, int bytes) +static inline void ubd_finish(struct request *req, int error) { - if(bytes < 0){ - ubd_end_request(req, 0, 0); - return; - } - - ubd_end_request(req, bytes, 1); + spin_lock(&ubd_io_lock); + __ubd_finish(req, error); + spin_unlock(&ubd_io_lock); } -static inline void ubd_finish(struct request *req, int bytes) +/* Called without ubd_io_lock held */ +static void ubd_handler(void) { - spin_lock(&ubd_io_lock); - __ubd_finish(req, bytes); - spin_unlock(&ubd_io_lock); + struct io_thread_req req; + struct request *rq = elv_next_request(ubd_queue); + int n; + + do_ubd = NULL; + intr_count++; + n = os_read_file(thread_fd, &req, sizeof(req)); + if(n != sizeof(req)){ + printk(KERN_ERR "Pid %d - spurious interrupt in ubd_handler, " + "err = %d\n", os_getpid(), -n); + spin_lock(&ubd_io_lock); + end_request(rq, 0); + spin_unlock(&ubd_io_lock); + return; + } + + ubd_finish(rq, req.error); + reactivate_fd(thread_fd, UBD_IRQ); + do_ubd_request(ubd_queue); } -struct bitmap_io { - atomic_t count; - struct aio_context aio; -}; - -struct ubd_aio { - struct aio_context aio; - struct request *req; - int len; - struct bitmap_io *bitmap; - void *bitmap_buf; -}; - -static int ubd_reply_fd = -1; - static irqreturn_t ubd_intr(int irq, void *dev, struct pt_regs *unused) { - struct aio_thread_reply reply; - struct ubd_aio *aio; - struct request *req; - int err, n, fd = (int) (long) dev; - - while(1){ - err = os_read_file(fd, &reply, sizeof(reply)); - if(err == -EAGAIN) - break; - if(err < 0){ - printk("ubd_aio_handler - read returned err %d\n", - -err); - break; - } - - aio = container_of(reply.data, struct ubd_aio, aio); - n = reply.err; - - if(n == 0){ - req = aio->req; - req->nr_sectors -= aio->len >> 9; - - if((aio->bitmap != NULL) && - (atomic_dec_and_test(&aio->bitmap->count))){ - aio->aio = aio->bitmap->aio; - aio->len = 0; - kfree(aio->bitmap); - aio->bitmap = NULL; - submit_aio(&aio->aio); - } - else { - if((req->nr_sectors == 0) && - (aio->bitmap == NULL)){ - int len = req->hard_nr_sectors << 9; - ubd_finish(req, len); - } - - if(aio->bitmap_buf != NULL) - kfree(aio->bitmap_buf); - kfree(aio); - } - } - else if(n < 0){ - ubd_finish(aio->req, n); - if(aio->bitmap != NULL) - kfree(aio->bitmap); - if(aio->bitmap_buf != NULL) - kfree(aio->bitmap_buf); - kfree(aio); - } - } - reactivate_fd(fd, UBD_IRQ); + ubd_handler(); + return(IRQ_HANDLED); +} - do_ubd_request(ubd_queue); +/* Only changed by ubd_init, which is an initcall. */ +static int io_pid = -1; - return(IRQ_HANDLED); +void kill_io_thread(void) +{ + if(io_pid != -1) + os_kill_process(io_pid, 1); } +__uml_exitcall(kill_io_thread); + static int ubd_file_size(struct ubd *dev, __u64 *size_out) { char *file; @@ -608,7 +569,7 @@ static int ubd_open_dev(struct ubd *dev) &dev->cow.data_offset, create_ptr); if((dev->fd == -ENOENT) && create_cow){ - dev->fd = create_cow_file(dev->file, dev->cow.file, + dev->fd = create_cow_file(dev->file, dev->cow.file, dev->openflags, 1 << 9, PAGE_SIZE, &dev->cow.bitmap_offset, &dev->cow.bitmap_len, @@ -870,10 +831,6 @@ int ubd_init(void) { int i; - ubd_reply_fd = init_aio_irq(UBD_IRQ, "ubd", ubd_intr); - if(ubd_reply_fd < 0) - printk("Setting up ubd AIO failed, err = %d\n", ubd_reply_fd); - devfs_mk_dir("ubd"); if (register_blkdev(MAJOR_NR, "ubd")) return -1; @@ -884,7 +841,6 @@ int ubd_init(void) return -1; } - blk_queue_max_hw_segments(ubd_queue, MAX_SG); if (fake_major != MAJOR_NR) { char name[sizeof("ubd_nnn\0")]; @@ -896,12 +852,40 @@ int ubd_init(void) driver_register(&ubd_driver); for (i = 0; i < MAX_DEV; i++) ubd_add(i); - return 0; } late_initcall(ubd_init); +int ubd_driver_init(void){ + unsigned long stack; + int err; + + /* Set by CONFIG_BLK_DEV_UBD_SYNC or ubd=sync.*/ + if(global_openflags.s){ + printk(KERN_INFO "ubd: Synchronous mode\n"); + /* Letting ubd=sync be like using ubd#s= instead of ubd#= is + * enough. So use anyway the io thread. */ + } + stack = alloc_stack(0, 0); + io_pid = start_io_thread(stack + PAGE_SIZE - sizeof(void *), + &thread_fd); + if(io_pid < 0){ + printk(KERN_ERR + "ubd : Failed to start I/O thread (errno = %d) - " + "falling back to synchronous I/O\n", -io_pid); + io_pid = -1; + return(0); + } + err = um_request_irq(UBD_IRQ, thread_fd, IRQ_READ, ubd_intr, + SA_INTERRUPT, "ubd", ubd_dev); + if(err != 0) + printk(KERN_ERR "um_request_irq failed - errno = %d\n", -err); + return(err); +} + +device_initcall(ubd_driver_init); + static int ubd_open(struct inode *inode, struct file *filp) { struct gendisk *disk = inode->i_bdev->bd_disk; @@ -939,55 +923,105 @@ static int ubd_release(struct inode * inode, struct file * file) return(0); } -static void cowify_bitmap(struct io_thread_req *req, unsigned long *bitmap) +static void cowify_bitmap(__u64 io_offset, int length, unsigned long *cow_mask, + __u64 *cow_offset, unsigned long *bitmap, + __u64 bitmap_offset, unsigned long *bitmap_words, + __u64 bitmap_len) { - __u64 sector = req->offset / req->sectorsize; - int i; + __u64 sector = io_offset >> 9; + int i, update_bitmap = 0; + + for(i = 0; i < length >> 9; i++){ + if(cow_mask != NULL) + ubd_set_bit(i, (unsigned char *) cow_mask); + if(ubd_test_bit(sector + i, (unsigned char *) bitmap)) + continue; - for(i = 0; i < req->length / req->sectorsize; i++){ - if(ubd_test_bit(sector + i, bitmap)) - continue; + update_bitmap = 1; + ubd_set_bit(sector + i, (unsigned char *) bitmap); + } + + if(!update_bitmap) + return; - if(req->bitmap_start == -1) - req->bitmap_start = sector + i; - req->bitmap_end = sector + i + 1; + *cow_offset = sector / (sizeof(unsigned long) * 8); - ubd_set_bit(sector + i, bitmap); - } + /* This takes care of the case where we're exactly at the end of the + * device, and *cow_offset + 1 is off the end. So, just back it up + * by one word. Thanks to Lynn Kerby for the fix and James McMechan + * for the original diagnosis. + */ + if(*cow_offset == ((bitmap_len + sizeof(unsigned long) - 1) / + sizeof(unsigned long) - 1)) + (*cow_offset)--; + + bitmap_words[0] = bitmap[*cow_offset]; + bitmap_words[1] = bitmap[*cow_offset + 1]; + + *cow_offset *= sizeof(unsigned long); + *cow_offset += bitmap_offset; +} + +static void cowify_req(struct io_thread_req *req, unsigned long *bitmap, + __u64 bitmap_offset, __u64 bitmap_len) +{ + __u64 sector = req->offset >> 9; + int i; + + if(req->length > (sizeof(req->sector_mask) * 8) << 9) + panic("Operation too long"); + + if(req->op == UBD_READ) { + for(i = 0; i < req->length >> 9; i++){ + if(ubd_test_bit(sector + i, (unsigned char *) bitmap)) + ubd_set_bit(i, (unsigned char *) + &req->sector_mask); + } + } + else cowify_bitmap(req->offset, req->length, &req->sector_mask, + &req->cow_offset, bitmap, bitmap_offset, + req->bitmap_words, bitmap_len); } /* Called with ubd_io_lock held */ -static int prepare_request(struct request *req, struct io_thread_req *io_req, - unsigned long long offset, int page_offset, - int len, struct page *page) +static int prepare_request(struct request *req, struct io_thread_req *io_req) { struct gendisk *disk = req->rq_disk; struct ubd *dev = disk->private_data; + __u64 offset; + int len; + + if(req->rq_status == RQ_INACTIVE) return(1); /* This should be impossible now */ if((rq_data_dir(req) == WRITE) && !dev->openflags.w){ printk("Write attempted on readonly ubd device %s\n", disk->disk_name); - ubd_end_request(req, 0, 0); + end_request(req, 0); return(1); } + offset = ((__u64) req->sector) << 9; + len = req->current_nr_sectors << 9; + io_req->fds[0] = (dev->cow.file != NULL) ? dev->cow.fd : dev->fd; io_req->fds[1] = dev->fd; + io_req->cow_offset = -1; io_req->offset = offset; io_req->length = len; io_req->error = 0; - io_req->op = (rq_data_dir(req) == READ) ? AIO_READ : AIO_WRITE; + io_req->sector_mask = 0; + + io_req->op = (rq_data_dir(req) == READ) ? UBD_READ : UBD_WRITE; io_req->offsets[0] = 0; io_req->offsets[1] = dev->cow.data_offset; - io_req->buffer = page_address(page) + page_offset; + io_req->buffer = req->buffer; io_req->sectorsize = 1 << 9; - io_req->bitmap_offset = dev->cow.bitmap_offset; - io_req->bitmap_start = -1; - io_req->bitmap_end = -1; - if((dev->cow.file != NULL) && (io_req->op == UBD_WRITE)) - cowify_bitmap(io_req, dev->cow.bitmap); + if(dev->cow.file != NULL) + cowify_req(io_req, dev->cow.bitmap, dev->cow.bitmap_offset, + dev->cow.bitmap_len); + return(0); } @@ -996,36 +1030,30 @@ static void do_ubd_request(request_queue_t *q) { struct io_thread_req io_req; struct request *req; - __u64 sector; - int err; - - if(in_ubd) - return; - in_ubd = 1; - while((req = elv_next_request(q)) != NULL){ - struct gendisk *disk = req->rq_disk; - struct ubd *dev = disk->private_data; - int n, i; - - blkdev_dequeue_request(req); - - sector = req->sector; - n = blk_rq_map_sg(q, req, dev->sg); - - for(i = 0; i < n; i++){ - struct scatterlist *sg = &dev->sg[i]; - - err = prepare_request(req, &io_req, sector << 9, - sg->offset, sg->length, - sg->page); - if(err) - continue; - - sector += sg->length >> 9; - do_io(&io_req, req, dev->cow.bitmap); + int err, n; + + if(thread_fd == -1){ + while((req = elv_next_request(q)) != NULL){ + err = prepare_request(req, &io_req); + if(!err){ + do_io(&io_req); + __ubd_finish(req, io_req.error); + } + } + } + else { + if(do_ubd || (req = elv_next_request(q)) == NULL) + return; + err = prepare_request(req, &io_req); + if(!err){ + do_ubd = ubd_handler; + n = os_write_file(thread_fd, (char *) &io_req, + sizeof(io_req)); + if(n != sizeof(io_req)) + printk("write to io thread failed, " + "errno = %d\n", -n); } } - in_ubd = 0; } static int ubd_ioctl(struct inode * inode, struct file * file, @@ -1241,95 +1269,131 @@ int create_cow_file(char *cow_file, char *backing_file, struct openflags flags, return(err); } -void do_io(struct io_thread_req *req, struct request *r, unsigned long *bitmap) +static int update_bitmap(struct io_thread_req *req) { - struct ubd_aio *aio; - struct bitmap_io *bitmap_io = NULL; - char *buf; - void *bitmap_buf = NULL; - unsigned long len, sector; - int nsectors, start, end, bit, err; - __u64 off; - - if(req->bitmap_start != -1){ - /* Round up to the nearest word */ - int round = sizeof(unsigned long); - len = (req->bitmap_end - req->bitmap_start + - round * 8 - 1) / (round * 8); - len *= round; - - off = req->bitmap_start / (8 * round); - off *= round; - - bitmap_io = kmalloc(sizeof(*bitmap_io), GFP_KERNEL); - if(bitmap_io == NULL){ - printk("Failed to kmalloc bitmap IO\n"); - req->error = 1; - return; - } + int n; - bitmap_buf = kmalloc(len, GFP_KERNEL); - if(bitmap_buf == NULL){ - printk("do_io : kmalloc of bitmap chunk " - "failed\n"); - kfree(bitmap_io); - req->error = 1; - return; - } - memcpy(bitmap_buf, &bitmap[off / sizeof(bitmap[0])], len); - - *bitmap_io = ((struct bitmap_io) - { .count = ATOMIC_INIT(0), - .aio = INIT_AIO(AIO_WRITE, req->fds[1], - bitmap_buf, len, - req->bitmap_offset + off, - ubd_reply_fd) } ); - } + if(req->cow_offset == -1) + return(0); - nsectors = req->length / req->sectorsize; - start = 0; - end = nsectors; - bit = 0; - do { - if(bitmap != NULL){ - sector = req->offset / req->sectorsize; - bit = ubd_test_bit(sector + start, bitmap); - end = start; - while((end < nsectors) && - (ubd_test_bit(sector + end, bitmap) == bit)) - end++; - } + n = os_seek_file(req->fds[1], req->cow_offset); + if(n < 0){ + printk("do_io - bitmap lseek failed : err = %d\n", -n); + return(1); + } - off = req->offsets[bit] + req->offset + - start * req->sectorsize; - len = (end - start) * req->sectorsize; - buf = &req->buffer[start * req->sectorsize]; + n = os_write_file(req->fds[1], &req->bitmap_words, + sizeof(req->bitmap_words)); + if(n != sizeof(req->bitmap_words)){ + printk("do_io - bitmap update failed, err = %d fd = %d\n", -n, + req->fds[1]); + return(1); + } - aio = kmalloc(sizeof(*aio), GFP_KERNEL); - if(aio == NULL){ - req->error = 1; - return; - } + return(0); +} - *aio = ((struct ubd_aio) - { .aio = INIT_AIO(req->op, req->fds[bit], buf, - len, off, ubd_reply_fd), - .len = len, - .req = r, - .bitmap = bitmap_io, - .bitmap_buf = bitmap_buf }); - - if(aio->bitmap != NULL) - atomic_inc(&aio->bitmap->count); - - err = submit_aio(&aio->aio); - if(err){ - printk("do_io - submit_aio failed, " - "err = %d\n", err); - req->error = 1; - return; - } +void do_io(struct io_thread_req *req) +{ + char *buf; + unsigned long len; + int n, nsectors, start, end, bit; + int err; + __u64 off; + + nsectors = req->length / req->sectorsize; + start = 0; + do { + bit = ubd_test_bit(start, (unsigned char *) &req->sector_mask); + end = start; + while((end < nsectors) && + (ubd_test_bit(end, (unsigned char *) + &req->sector_mask) == bit)) + end++; + + off = req->offset + req->offsets[bit] + + start * req->sectorsize; + len = (end - start) * req->sectorsize; + buf = &req->buffer[start * req->sectorsize]; + + err = os_seek_file(req->fds[bit], off); + if(err < 0){ + printk("do_io - lseek failed : err = %d\n", -err); + req->error = 1; + return; + } + if(req->op == UBD_READ){ + n = 0; + do { + buf = &buf[n]; + len -= n; + n = os_read_file(req->fds[bit], buf, len); + if (n < 0) { + printk("do_io - read failed, err = %d " + "fd = %d\n", -n, req->fds[bit]); + req->error = 1; + return; + } + } while((n < len) && (n != 0)); + if (n < len) memset(&buf[n], 0, len - n); + } else { + n = os_write_file(req->fds[bit], buf, len); + if(n != len){ + printk("do_io - write failed err = %d " + "fd = %d\n", -n, req->fds[bit]); + req->error = 1; + return; + } + } + + start = end; + } while(start < nsectors); - start = end; - } while(start < nsectors); + req->error = update_bitmap(req); } + +/* Changed in start_io_thread, which is serialized by being called only + * from ubd_init, which is an initcall. + */ +int kernel_fd = -1; + +/* Only changed by the io thread */ +int io_count = 0; + +int io_thread(void *arg) +{ + struct io_thread_req req; + int n; + + ignore_sigwinch_sig(); + while(1){ + n = os_read_file(kernel_fd, &req, sizeof(req)); + if(n != sizeof(req)){ + if(n < 0) + printk("io_thread - read failed, fd = %d, " + "err = %d\n", kernel_fd, -n); + else { + printk("io_thread - short read, fd = %d, " + "length = %d\n", kernel_fd, n); + } + continue; + } + io_count++; + do_io(&req); + n = os_write_file(kernel_fd, &req, sizeof(req)); + if(n != sizeof(req)) + printk("io_thread - write failed, fd = %d, err = %d\n", + kernel_fd, -n); + } +} + +/* + * Overrides for Emacs so that we follow Linus's tabbing style. + * Emacs will notice this stuff at the end of the file and automatically + * adjust the settings for this buffer only. This must remain at the end + * of the file. + * --------------------------------------------------------------------------- + * Local variables: + * c-file-style: "linux" + * End: + */ diff --git a/arch/um/drivers/ubd_user.c b/arch/um/drivers/ubd_user.c new file mode 100644 index 00000000000..b94d2bc4fe0 --- /dev/null +++ b/arch/um/drivers/ubd_user.c @@ -0,0 +1,75 @@ +/* + * Copyright (C) 2000, 2001, 2002 Jeff Dike (jdike@karaya.com) + * Copyright (C) 2001 Ridgerun,Inc (glonnon@ridgerun.com) + * Licensed under the GPL + */ + +#include <stddef.h> +#include <unistd.h> +#include <errno.h> +#include <sched.h> +#include <signal.h> +#include <string.h> +#include <netinet/in.h> +#include <sys/time.h> +#include <sys/socket.h> +#include <sys/mman.h> +#include <sys/param.h> +#include "asm/types.h" +#include "user_util.h" +#include "kern_util.h" +#include "user.h" +#include "ubd_user.h" +#include "os.h" +#include "cow.h" + +#include <endian.h> +#include <byteswap.h> + +void ignore_sigwinch_sig(void) +{ + signal(SIGWINCH, SIG_IGN); +} + +int start_io_thread(unsigned long sp, int *fd_out) +{ + int pid, fds[2], err; + + err = os_pipe(fds, 1, 1); + if(err < 0){ + printk("start_io_thread - os_pipe failed, err = %d\n", -err); + goto out; + } + + kernel_fd = fds[0]; + *fd_out = fds[1]; + + pid = clone(io_thread, (void *) sp, CLONE_FILES | CLONE_VM | SIGCHLD, + NULL); + if(pid < 0){ + printk("start_io_thread - clone failed : errno = %d\n", errno); + err = -errno; + goto out_close; + } + + return(pid); + + out_close: + os_close_file(fds[0]); + os_close_file(fds[1]); + kernel_fd = -1; + *fd_out = -1; + out: + return(err); +} + +/* + * Overrides for Emacs so that we follow Linus's tabbing style. + * Emacs will notice this stuff at the end of the file and automatically + * adjust the settings for this buffer only. This must remain at the end + * of the file. + * --------------------------------------------------------------------------- + * Local variables: + * c-file-style: "linux" + * End: + */ diff --git a/arch/um/include/aio.h b/arch/um/include/aio.h index 83f16877ab0..423bae9153f 100644 --- a/arch/um/include/aio.h +++ b/arch/um/include/aio.h @@ -14,27 +14,15 @@ struct aio_thread_reply { }; struct aio_context { - enum aio_type type; - int fd; - void *data; - int len; - unsigned long long offset; int reply_fd; struct aio_context *next; }; -#define INIT_AIO(aio_type, aio_fd, aio_data, aio_len, aio_offset, \ - aio_reply_fd) \ - { .type = aio_type, \ - .fd = aio_fd, \ - .data = aio_data, \ - .len = aio_len, \ - .offset = aio_offset, \ - .reply_fd = aio_reply_fd } - #define INIT_AIO_CONTEXT { .reply_fd = -1, \ .next = NULL } -extern int submit_aio(struct aio_context *aio); +extern int submit_aio(enum aio_type type, int fd, char *buf, int len, + unsigned long long offset, int reply_fd, + struct aio_context *aio); #endif diff --git a/arch/um/include/os.h b/arch/um/include/os.h index 6f766e1faec..2e58e304b8b 100644 --- a/arch/um/include/os.h +++ b/arch/um/include/os.h @@ -6,6 +6,7 @@ #ifndef __OS_H__ #define __OS_H__ +#include "uml-config.h" #include "asm/types.h" #include "../os/include/file.h" @@ -159,7 +160,11 @@ extern int can_do_skas(void); /* Make sure they are clear when running in TT mode. Required by * SEGV_MAYBE_FIXABLE */ +#ifdef UML_CONFIG_MODE_SKAS #define clear_can_do_skas() do { ptrace_faultinfo = proc_mm = 0; } while (0) +#else +#define clear_can_do_skas() do {} while (0) +#endif /* mem.c */ extern int create_mem_file(unsigned long len); diff --git a/arch/um/os-Linux/aio.c b/arch/um/os-Linux/aio.c index f6e64026f99..41cfb094420 100644 --- a/arch/um/os-Linux/aio.c +++ b/arch/um/os-Linux/aio.c @@ -6,7 +6,6 @@ #include <stdlib.h> #include <unistd.h> #include <signal.h> -#include <string.h> #include <errno.h> #include <sched.h> #include <sys/syscall.h> @@ -17,31 +16,18 @@ #include "user.h" #include "mode.h" +struct aio_thread_req { + enum aio_type type; + int io_fd; + unsigned long long offset; + char *buf; + int len; + struct aio_context *aio; +}; + static int aio_req_fd_r = -1; static int aio_req_fd_w = -1; -static int update_aio(struct aio_context *aio, int res) -{ - if(res < 0) - aio->len = res; - else if((res == 0) && (aio->type == AIO_READ)){ - /* This is the EOF case - we have hit the end of the file - * and it ends in a partial block, so we fill the end of - * the block with zeros and claim success. - */ - memset(aio->data, 0, aio->len); - aio->len = 0; - } - else if(res > 0){ - aio->len -= res; - aio->data += res; - aio->offset += res; - return aio->len; - } - - return 0; -} - #if defined(HAVE_AIO_ABI) #include <linux/aio_abi.h> @@ -80,7 +66,8 @@ static long io_getevents(aio_context_t ctx_id, long min_nr, long nr, * that it now backs the mmapped area. */ -static int do_aio(aio_context_t ctx, struct aio_context *aio) +static int do_aio(aio_context_t ctx, enum aio_type type, int fd, char *buf, + int len, unsigned long long offset, struct aio_context *aio) { struct iocb iocb, *iocbp = &iocb; char c; @@ -88,39 +75,40 @@ static int do_aio(aio_context_t ctx, struct aio_context *aio) iocb = ((struct iocb) { .aio_data = (unsigned long) aio, .aio_reqprio = 0, - .aio_fildes = aio->fd, - .aio_buf = (unsigned long) aio->data, - .aio_nbytes = aio->len, - .aio_offset = aio->offset, + .aio_fildes = fd, + .aio_buf = (unsigned long) buf, + .aio_nbytes = len, + .aio_offset = offset, .aio_reserved1 = 0, .aio_reserved2 = 0, .aio_reserved3 = 0 }); - switch(aio->type){ + switch(type){ case AIO_READ: iocb.aio_lio_opcode = IOCB_CMD_PREAD; + err = io_submit(ctx, 1, &iocbp); break; case AIO_WRITE: iocb.aio_lio_opcode = IOCB_CMD_PWRITE; + err = io_submit(ctx, 1, &iocbp); break; case AIO_MMAP: iocb.aio_lio_opcode = IOCB_CMD_PREAD; iocb.aio_buf = (unsigned long) &c; iocb.aio_nbytes = sizeof(c); + err = io_submit(ctx, 1, &iocbp); break; default: - printk("Bogus op in do_aio - %d\n", aio->type); + printk("Bogus op in do_aio - %d\n", type); err = -EINVAL; - goto out; + break; } - err = io_submit(ctx, 1, &iocbp); if(err > 0) err = 0; else err = -errno; - out: return err; } @@ -129,9 +117,8 @@ static aio_context_t ctx = 0; static int aio_thread(void *arg) { struct aio_thread_reply reply; - struct aio_context *aio; struct io_event event; - int err, n; + int err, n, reply_fd; signal(SIGWINCH, SIG_IGN); @@ -144,22 +131,14 @@ static int aio_thread(void *arg) "errno = %d\n", errno); } else { - /* This is safe as we've just a pointer here. */ - aio = (struct aio_context *) (long) event.data; - if(update_aio(aio, event.res)){ - do_aio(ctx, aio); - continue; - } - reply = ((struct aio_thread_reply) - { .data = aio, - .err = aio->len }); - err = os_write_file(aio->reply_fd, &reply, - sizeof(reply)); + { .data = (void *) (long) event.data, + .err = event.res }); + reply_fd = ((struct aio_context *) reply.data)->reply_fd; + err = os_write_file(reply_fd, &reply, sizeof(reply)); if(err != sizeof(reply)) - printk("aio_thread - write failed, " - "fd = %d, err = %d\n", aio->reply_fd, - -err); + printk("aio_thread - write failed, fd = %d, " + "err = %d\n", aio_req_fd_r, -err); } } return 0; @@ -167,35 +146,35 @@ static int aio_thread(void *arg) #endif -static int do_not_aio(struct aio_context *aio) +static int do_not_aio(struct aio_thread_req *req) { char c; int err; - switch(aio->type){ + switch(req->type){ case AIO_READ: - err = os_seek_file(aio->fd, aio->offset); + err = os_seek_file(req->io_fd, req->offset); if(err) goto out; - err = os_read_file(aio->fd, aio->data, aio->len); + err = os_read_file(req->io_fd, req->buf, req->len); break; case AIO_WRITE: - err = os_seek_file(aio->fd, aio->offset); + err = os_seek_file(req->io_fd, req->offset); if(err) goto out; - err = os_write_file(aio->fd, aio->data, aio->len); + err = os_write_file(req->io_fd, req->buf, req->len); break; case AIO_MMAP: - err = os_seek_file(aio->fd, aio->offset); + err = os_seek_file(req->io_fd, req->offset); if(err) goto out; - err = os_read_file(aio->fd, &c, sizeof(c)); + err = os_read_file(req->io_fd, &c, sizeof(c)); break; default: - printk("do_not_aio - bad request type : %d\n", aio->type); + printk("do_not_aio - bad request type : %d\n", req->type); err = -EINVAL; break; } @@ -206,14 +185,14 @@ static int do_not_aio(struct aio_context *aio) static int not_aio_thread(void *arg) { - struct aio_context *aio; + struct aio_thread_req req; struct aio_thread_reply reply; int err; signal(SIGWINCH, SIG_IGN); while(1){ - err = os_read_file(aio_req_fd_r, &aio, sizeof(aio)); - if(err != sizeof(aio)){ + err = os_read_file(aio_req_fd_r, &req, sizeof(req)); + if(err != sizeof(req)){ if(err < 0) printk("not_aio_thread - read failed, " "fd = %d, err = %d\n", aio_req_fd_r, @@ -224,34 +203,17 @@ static int not_aio_thread(void *arg) } continue; } - again: - err = do_not_aio(aio); - - if(update_aio(aio, err)) - goto again; - - reply = ((struct aio_thread_reply) { .data = aio, - .err = aio->len }); - err = os_write_file(aio->reply_fd, &reply, sizeof(reply)); + err = do_not_aio(&req); + reply = ((struct aio_thread_reply) { .data = req.aio, + .err = err }); + err = os_write_file(req.aio->reply_fd, &reply, sizeof(reply)); if(err != sizeof(reply)) printk("not_aio_thread - write failed, fd = %d, " "err = %d\n", aio_req_fd_r, -err); } } -static int submit_aio_24(struct aio_context *aio) -{ - int err; - - err = os_write_file(aio_req_fd_w, &aio, sizeof(aio)); - if(err == sizeof(aio)) - err = 0; - - return err; -} - static int aio_pid = -1; -static int (*submit_proc)(struct aio_context *aio); static int init_aio_24(void) { @@ -283,33 +245,11 @@ static int init_aio_24(void) #endif printk("2.6 host AIO support not used - falling back to I/O " "thread\n"); - - submit_proc = submit_aio_24; - return 0; } #ifdef HAVE_AIO_ABI #define DEFAULT_24_AIO 0 -static int submit_aio_26(struct aio_context *aio) -{ - struct aio_thread_reply reply; - int err; - - err = do_aio(ctx, aio); - if(err){ - reply = ((struct aio_thread_reply) { .data = aio, - .err = err }); - err = os_write_file(aio->reply_fd, &reply, sizeof(reply)); - if(err != sizeof(reply)) - printk("submit_aio_26 - write failed, " - "fd = %d, err = %d\n", aio->reply_fd, -err); - else err = 0; - } - - return err; -} - static int init_aio_26(void) { unsigned long stack; @@ -330,22 +270,39 @@ static int init_aio_26(void) aio_pid = err; printk("Using 2.6 host AIO\n"); + return 0; +} + +static int submit_aio_26(enum aio_type type, int io_fd, char *buf, int len, + unsigned long long offset, struct aio_context *aio) +{ + struct aio_thread_reply reply; + int err; - submit_proc = submit_aio_26; + err = do_aio(ctx, type, io_fd, buf, len, offset, aio); + if(err){ + reply = ((struct aio_thread_reply) { .data = aio, + .err = err }); + err = os_write_file(aio->reply_fd, &reply, sizeof(reply)); + if(err != sizeof(reply)) + printk("submit_aio_26 - write failed, " + "fd = %d, err = %d\n", aio->reply_fd, -err); + else err = 0; + } - return 0; + return err; } #else #define DEFAULT_24_AIO 1 -static int submit_aio_26(struct aio_context *aio) +static int init_aio_26(void) { return -ENOSYS; } -static int init_aio_26(void) +static int submit_aio_26(enum aio_type type, int io_fd, char *buf, int len, + unsigned long long offset, struct aio_context *aio) { - submit_proc = submit_aio_26; return -ENOSYS; } #endif @@ -412,7 +369,33 @@ static void exit_aio(void) __uml_exitcall(exit_aio); -int submit_aio(struct aio_context *aio) +static int submit_aio_24(enum aio_type type, int io_fd, char *buf, int len, + unsigned long long offset, struct aio_context *aio) { - return (*submit_proc)(aio); + struct aio_thread_req req = { .type = type, + .io_fd = io_fd, + .offset = offset, + .buf = buf, + .len = len, + .aio = aio, + }; + int err; + + err = os_write_file(aio_req_fd_w, &req, sizeof(req)); + if(err == sizeof(req)) + err = 0; + + return err; +} + +int submit_aio(enum aio_type type, int io_fd, char *buf, int len, + unsigned long long offset, int reply_fd, + struct aio_context *aio) +{ + aio->reply_fd = reply_fd; + if(aio_24) + return submit_aio_24(type, io_fd, buf, len, offset, aio); + else { + return submit_aio_26(type, io_fd, buf, len, offset, aio); + } } diff --git a/drivers/net/e100.c b/drivers/net/e100.c index fbf1c06ec5c..40887f09b68 100644 --- a/drivers/net/e100.c +++ b/drivers/net/e100.c @@ -903,8 +903,8 @@ static void mdio_write(struct net_device *netdev, int addr, int reg, int data) static void e100_get_defaults(struct nic *nic) { - struct param_range rfds = { .min = 16, .max = 256, .count = 256 }; - struct param_range cbs = { .min = 64, .max = 256, .count = 128 }; + struct param_range rfds = { .min = 16, .max = 256, .count = 64 }; + struct param_range cbs = { .min = 64, .max = 256, .count = 64 }; pci_read_config_byte(nic->pdev, PCI_REVISION_ID, &nic->rev_id); /* MAC type is encoded as rev ID; exception: ICH is treated as 82559 */ @@ -1007,213 +1007,25 @@ static void e100_configure(struct nic *nic, struct cb *cb, struct sk_buff *skb) c[16], c[17], c[18], c[19], c[20], c[21], c[22], c[23]); } -/********************************************************/ -/* Micro code for 8086:1229 Rev 8 */ -/********************************************************/ - -/* Parameter values for the D101M B-step */ -#define D101M_CPUSAVER_TIMER_DWORD 78 -#define D101M_CPUSAVER_BUNDLE_DWORD 65 -#define D101M_CPUSAVER_MIN_SIZE_DWORD 126 - -#define D101M_B_RCVBUNDLE_UCODE \ -{\ -0x00550215, 0xFFFF0437, 0xFFFFFFFF, 0x06A70789, 0xFFFFFFFF, 0x0558FFFF, \ -0x000C0001, 0x00101312, 0x000C0008, 0x00380216, \ -0x0010009C, 0x00204056, 0x002380CC, 0x00380056, \ -0x0010009C, 0x00244C0B, 0x00000800, 0x00124818, \ -0x00380438, 0x00000000, 0x00140000, 0x00380555, \ -0x00308000, 0x00100662, 0x00100561, 0x000E0408, \ -0x00134861, 0x000C0002, 0x00103093, 0x00308000, \ -0x00100624, 0x00100561, 0x000E0408, 0x00100861, \ -0x000C007E, 0x00222C21, 0x000C0002, 0x00103093, \ -0x00380C7A, 0x00080000, 0x00103090, 0x00380C7A, \ -0x00000000, 0x00000000, 0x00000000, 0x00000000, \ -0x0010009C, 0x00244C2D, 0x00010004, 0x00041000, \ -0x003A0437, 0x00044010, 0x0038078A, 0x00000000, \ -0x00100099, 0x00206C7A, 0x0010009C, 0x00244C48, \ -0x00130824, 0x000C0001, 0x00101213, 0x00260C75, \ -0x00041000, 0x00010004, 0x00130826, 0x000C0006, \ -0x002206A8, 0x0013C926, 0x00101313, 0x003806A8, \ -0x00000000, 0x00000000, 0x00000000, 0x00000000, \ -0x00000000, 0x00000000, 0x00000000, 0x00000000, \ -0x00080600, 0x00101B10, 0x00050004, 0x00100826, \ -0x00101210, 0x00380C34, 0x00000000, 0x00000000, \ -0x0021155B, 0x00100099, 0x00206559, 0x0010009C, \ -0x00244559, 0x00130836, 0x000C0000, 0x00220C62, \ -0x000C0001, 0x00101B13, 0x00229C0E, 0x00210C0E, \ -0x00226C0E, 0x00216C0E, 0x0022FC0E, 0x00215C0E, \ -0x00214C0E, 0x00380555, 0x00010004, 0x00041000, \ -0x00278C67, 0x00040800, 0x00018100, 0x003A0437, \ -0x00130826, 0x000C0001, 0x00220559, 0x00101313, \ -0x00380559, 0x00000000, 0x00000000, 0x00000000, \ -0x00000000, 0x00000000, 0x00000000, 0x00000000, \ -0x00000000, 0x00130831, 0x0010090B, 0x00124813, \ -0x000CFF80, 0x002606AB, 0x00041000, 0x00010004, \ -0x003806A8, 0x00000000, 0x00000000, 0x00000000, \ -} - -/********************************************************/ -/* Micro code for 8086:1229 Rev 9 */ -/********************************************************/ - -/* Parameter values for the D101S */ -#define D101S_CPUSAVER_TIMER_DWORD 78 -#define D101S_CPUSAVER_BUNDLE_DWORD 67 -#define D101S_CPUSAVER_MIN_SIZE_DWORD 128 - -#define D101S_RCVBUNDLE_UCODE \ -{\ -0x00550242, 0xFFFF047E, 0xFFFFFFFF, 0x06FF0818, 0xFFFFFFFF, 0x05A6FFFF, \ -0x000C0001, 0x00101312, 0x000C0008, 0x00380243, \ -0x0010009C, 0x00204056, 0x002380D0, 0x00380056, \ -0x0010009C, 0x00244F8B, 0x00000800, 0x00124818, \ -0x0038047F, 0x00000000, 0x00140000, 0x003805A3, \ -0x00308000, 0x00100610, 0x00100561, 0x000E0408, \ -0x00134861, 0x000C0002, 0x00103093, 0x00308000, \ -0x00100624, 0x00100561, 0x000E0408, 0x00100861, \ -0x000C007E, 0x00222FA1, 0x000C0002, 0x00103093, \ -0x00380F90, 0x00080000, 0x00103090, 0x00380F90, \ -0x00000000, 0x00000000, 0x00000000, 0x00000000, \ -0x0010009C, 0x00244FAD, 0x00010004, 0x00041000, \ -0x003A047E, 0x00044010, 0x00380819, 0x00000000, \ -0x00100099, 0x00206FFD, 0x0010009A, 0x0020AFFD, \ -0x0010009C, 0x00244FC8, 0x00130824, 0x000C0001, \ -0x00101213, 0x00260FF7, 0x00041000, 0x00010004, \ -0x00130826, 0x000C0006, 0x00220700, 0x0013C926, \ -0x00101313, 0x00380700, 0x00000000, 0x00000000, \ -0x00000000, 0x00000000, 0x00000000, 0x00000000, \ -0x00080600, 0x00101B10, 0x00050004, 0x00100826, \ -0x00101210, 0x00380FB6, 0x00000000, 0x00000000, \ -0x002115A9, 0x00100099, 0x002065A7, 0x0010009A, \ -0x0020A5A7, 0x0010009C, 0x002445A7, 0x00130836, \ -0x000C0000, 0x00220FE4, 0x000C0001, 0x00101B13, \ -0x00229F8E, 0x00210F8E, 0x00226F8E, 0x00216F8E, \ -0x0022FF8E, 0x00215F8E, 0x00214F8E, 0x003805A3, \ -0x00010004, 0x00041000, 0x00278FE9, 0x00040800, \ -0x00018100, 0x003A047E, 0x00130826, 0x000C0001, \ -0x002205A7, 0x00101313, 0x003805A7, 0x00000000, \ -0x00000000, 0x00000000, 0x00000000, 0x00000000, \ -0x00000000, 0x00000000, 0x00000000, 0x00130831, \ -0x0010090B, 0x00124813, 0x000CFF80, 0x00260703, \ -0x00041000, 0x00010004, 0x00380700 \ -} - -/********************************************************/ -/* Micro code for the 8086:1229 Rev F/10 */ -/********************************************************/ - -/* Parameter values for the D102 E-step */ -#define D102_E_CPUSAVER_TIMER_DWORD 42 -#define D102_E_CPUSAVER_BUNDLE_DWORD 54 -#define D102_E_CPUSAVER_MIN_SIZE_DWORD 46 - -#define D102_E_RCVBUNDLE_UCODE \ -{\ -0x007D028F, 0x0E4204F9, 0x14ED0C85, 0x14FA14E9, 0x0EF70E36, 0x1FFF1FFF, \ -0x00E014B9, 0x00000000, 0x00000000, 0x00000000, \ -0x00E014BD, 0x00000000, 0x00000000, 0x00000000, \ -0x00E014D5, 0x00000000, 0x00000000, 0x00000000, \ -0x00000000, 0x00000000, 0x00000000, 0x00000000, \ -0x00E014C1, 0x00000000, 0x00000000, 0x00000000, \ -0x00000000, 0x00000000, 0x00000000, 0x00000000, \ -0x00000000, 0x00000000, 0x00000000, 0x00000000, \ -0x00000000, 0x00000000, 0x00000000, 0x00000000, \ -0x00E014C8, 0x00000000, 0x00000000, 0x00000000, \ -0x00200600, 0x00E014EE, 0x00000000, 0x00000000, \ -0x0030FF80, 0x00940E46, 0x00038200, 0x00102000, \ -0x00E00E43, 0x00000000, 0x00000000, 0x00000000, \ -0x00300006, 0x00E014FB, 0x00000000, 0x00000000, \ -0x00000000, 0x00000000, 0x00000000, 0x00000000, \ -0x00000000, 0x00000000, 0x00000000, 0x00000000, \ -0x00000000, 0x00000000, 0x00000000, 0x00000000, \ -0x00906E41, 0x00800E3C, 0x00E00E39, 0x00000000, \ -0x00906EFD, 0x00900EFD, 0x00E00EF8, 0x00000000, \ -0x00000000, 0x00000000, 0x00000000, 0x00000000, \ -0x00000000, 0x00000000, 0x00000000, 0x00000000, \ -0x00000000, 0x00000000, 0x00000000, 0x00000000, \ -0x00000000, 0x00000000, 0x00000000, 0x00000000, \ -0x00000000, 0x00000000, 0x00000000, 0x00000000, \ -0x00000000, 0x00000000, 0x00000000, 0x00000000, \ -0x00000000, 0x00000000, 0x00000000, 0x00000000, \ -0x00000000, 0x00000000, 0x00000000, 0x00000000, \ -0x00000000, 0x00000000, 0x00000000, 0x00000000, \ -0x00000000, 0x00000000, 0x00000000, 0x00000000, \ -0x00000000, 0x00000000, 0x00000000, 0x00000000, \ -0x00000000, 0x00000000, 0x00000000, 0x00000000, \ -0x00000000, 0x00000000, 0x00000000, 0x00000000, \ -0x00000000, 0x00000000, 0x00000000, 0x00000000, \ -} - static void e100_load_ucode(struct nic *nic, struct cb *cb, struct sk_buff *skb) { -/* *INDENT-OFF* */ - static struct { - u32 ucode[UCODE_SIZE + 1]; - u8 mac; - u8 timer_dword; - u8 bundle_dword; - u8 min_size_dword; - } ucode_opts[] = { - { D101M_B_RCVBUNDLE_UCODE, - mac_82559_D101M, - D101M_CPUSAVER_TIMER_DWORD, - D101M_CPUSAVER_BUNDLE_DWORD, - D101M_CPUSAVER_MIN_SIZE_DWORD }, - { D101S_RCVBUNDLE_UCODE, - mac_82559_D101S, - D101S_CPUSAVER_TIMER_DWORD, - D101S_CPUSAVER_BUNDLE_DWORD, - D101S_CPUSAVER_MIN_SIZE_DWORD }, - { D102_E_RCVBUNDLE_UCODE, - mac_82551_F, - D102_E_CPUSAVER_TIMER_DWORD, - D102_E_CPUSAVER_BUNDLE_DWORD, - D102_E_CPUSAVER_MIN_SIZE_DWORD }, - { D102_E_RCVBUNDLE_UCODE, - mac_82551_10, - D102_E_CPUSAVER_TIMER_DWORD, - D102_E_CPUSAVER_BUNDLE_DWORD, - D102_E_CPUSAVER_MIN_SIZE_DWORD }, - { {0}, 0, 0, 0, 0} - }, *opts; -/* *INDENT-ON* */ - -#define BUNDLESMALL 1 -#define BUNDLEMAX 50 -#define INTDELAY 15000 - - opts = ucode_opts; - - /* do not load u-code for ICH devices */ - if (nic->flags & ich) - return; - - /* Search for ucode match against h/w rev_id */ - while (opts->mac) { - if (nic->mac == opts->mac) { - int i; - u32 *ucode = opts->ucode; - - /* Insert user-tunable settings */ - ucode[opts->timer_dword] &= 0xFFFF0000; - ucode[opts->timer_dword] |= - (u16) INTDELAY; - ucode[opts->bundle_dword] &= 0xFFFF0000; - ucode[opts->bundle_dword] |= (u16) BUNDLEMAX; - ucode[opts->min_size_dword] &= 0xFFFF0000; - ucode[opts->min_size_dword] |= - (BUNDLESMALL) ? 0xFFFF : 0xFF80; - - for(i = 0; i < UCODE_SIZE; i++) - cb->u.ucode[i] = cpu_to_le32(ucode[i]); - cb->command = cpu_to_le16(cb_ucode); - return; - } - opts++; - } + int i; + static const u32 ucode[UCODE_SIZE] = { + /* NFS packets are misinterpreted as TCO packets and + * incorrectly routed to the BMC over SMBus. This + * microcode patch checks the fragmented IP bit in the + * NFS/UDP header to distinguish between NFS and TCO. */ + 0x0EF70E36, 0x1FFF1FFF, 0x1FFF1FFF, 0x1FFF1FFF, 0x1FFF1FFF, + 0x1FFF1FFF, 0x00906E41, 0x00800E3C, 0x00E00E39, 0x00000000, + 0x00906EFD, 0x00900EFD, 0x00E00EF8, + }; - cb->command = cpu_to_le16(cb_nop); + if(nic->mac == mac_82551_F || nic->mac == mac_82551_10) { + for(i = 0; i < UCODE_SIZE; i++) + cb->u.ucode[i] = cpu_to_le32(ucode[i]); + cb->command = cpu_to_le16(cb_ucode); + } else + cb->command = cpu_to_le16(cb_nop); } static void e100_setup_iaaddr(struct nic *nic, struct cb *cb, diff --git a/drivers/scsi/sata_nv.c b/drivers/scsi/sata_nv.c index c05653c7779..cb832b03ec5 100644 --- a/drivers/scsi/sata_nv.c +++ b/drivers/scsi/sata_nv.c @@ -29,6 +29,8 @@ * NV-specific details such as register offsets, SATA phy location, * hotplug info, etc. * + * 0.09 + * - Fixed bug introduced by 0.08's MCP51 and MCP55 support. * * 0.08 * - Added support for MCP51 and MCP55. @@ -132,9 +134,7 @@ enum nv_host_type GENERIC, NFORCE2, NFORCE3, - CK804, - MCP51, - MCP55 + CK804 }; static struct pci_device_id nv_pci_tbl[] = { @@ -153,13 +153,13 @@ static struct pci_device_id nv_pci_tbl[] = { { PCI_VENDOR_ID_NVIDIA, PCI_DEVICE_ID_NVIDIA_NFORCE_MCP04_SATA2, PCI_ANY_ID, PCI_ANY_ID, 0, 0, CK804 }, { PCI_VENDOR_ID_NVIDIA, PCI_DEVICE_ID_NVIDIA_NFORCE_MCP51_SATA, - PCI_ANY_ID, PCI_ANY_ID, 0, 0, MCP51 }, + PCI_ANY_ID, PCI_ANY_ID, 0, 0, GENERIC }, { PCI_VENDOR_ID_NVIDIA, PCI_DEVICE_ID_NVIDIA_NFORCE_MCP51_SATA2, - PCI_ANY_ID, PCI_ANY_ID, 0, 0, MCP51 }, + PCI_ANY_ID, PCI_ANY_ID, 0, 0, GENERIC }, { PCI_VENDOR_ID_NVIDIA, PCI_DEVICE_ID_NVIDIA_NFORCE_MCP55_SATA, - PCI_ANY_ID, PCI_ANY_ID, 0, 0, MCP55 }, + PCI_ANY_ID, PCI_ANY_ID, 0, 0, GENERIC }, { PCI_VENDOR_ID_NVIDIA, PCI_DEVICE_ID_NVIDIA_NFORCE_MCP55_SATA2, - PCI_ANY_ID, PCI_ANY_ID, 0, 0, MCP55 }, + PCI_ANY_ID, PCI_ANY_ID, 0, 0, GENERIC }, { PCI_VENDOR_ID_NVIDIA, PCI_ANY_ID, PCI_ANY_ID, PCI_ANY_ID, PCI_CLASS_STORAGE_IDE<<8, 0xffff00, GENERIC }, diff --git a/include/asm-powerpc/timex.h b/include/asm-powerpc/timex.h index 51c5b316be5..c02d15aced9 100644 --- a/include/asm-powerpc/timex.h +++ b/include/asm-powerpc/timex.h @@ -10,7 +10,7 @@ #include <linux/config.h> #include <asm/cputable.h> -#define CLOCK_TICK_RATE 1193180 /* Underlying HZ */ +#define CLOCK_TICK_RATE 1024000 /* Underlying HZ */ typedef unsigned long cycles_t; diff --git a/include/asm-ppc/cputable.h b/include/asm-ppc/cputable.h index 41d8f8425c0..e17c492c870 100644 --- a/include/asm-ppc/cputable.h +++ b/include/asm-ppc/cputable.h @@ -24,6 +24,7 @@ #define PPC_FEATURE_HAS_SPE 0x00800000 #define PPC_FEATURE_HAS_EFP_SINGLE 0x00400000 #define PPC_FEATURE_HAS_EFP_DOUBLE 0x00200000 +#define PPC_FEATURE_NO_TB 0x00100000 #ifdef __KERNEL__ diff --git a/include/linux/netfilter/nfnetlink.h b/include/linux/netfilter/nfnetlink.h index 1d5b10ae239..f08e870100f 100644 --- a/include/linux/netfilter/nfnetlink.h +++ b/include/linux/netfilter/nfnetlink.h @@ -41,11 +41,15 @@ enum nfnetlink_groups { struct nfattr { u_int16_t nfa_len; - u_int16_t nfa_type; + u_int16_t nfa_type; /* we use 15 bits for the type, and the highest + * bit to indicate whether the payload is nested */ } __attribute__ ((packed)); -/* FIXME: Shamelessly copy and pasted from rtnetlink.h, it's time - * to put this in a generic file */ +/* FIXME: Apart from NFNL_NFA_NESTED shamelessly copy and pasted from + * rtnetlink.h, it's time to put this in a generic file */ + +#define NFNL_NFA_NEST 0x8000 +#define NFA_TYPE(attr) ((attr)->nfa_type & 0x7fff) #define NFA_ALIGNTO 4 #define NFA_ALIGN(len) (((len) + NFA_ALIGNTO - 1) & ~(NFA_ALIGNTO - 1)) @@ -59,7 +63,7 @@ struct nfattr #define NFA_PAYLOAD(nfa) ((int)((nfa)->nfa_len) - NFA_LENGTH(0)) #define NFA_NEST(skb, type) \ ({ struct nfattr *__start = (struct nfattr *) (skb)->tail; \ - NFA_PUT(skb, type, 0, NULL); \ + NFA_PUT(skb, (NFNL_NFA_NEST | type), 0, NULL); \ __start; }) #define NFA_NEST_END(skb, start) \ ({ (start)->nfa_len = ((skb)->tail - (unsigned char *) (start)); \ diff --git a/include/linux/netfilter/nfnetlink_conntrack.h b/include/linux/netfilter/nfnetlink_conntrack.h index 5c55751c78e..116fcaced90 100644 --- a/include/linux/netfilter/nfnetlink_conntrack.h +++ b/include/linux/netfilter/nfnetlink_conntrack.h @@ -70,15 +70,24 @@ enum ctattr_l4proto { enum ctattr_protoinfo { CTA_PROTOINFO_UNSPEC, - CTA_PROTOINFO_TCP_STATE, + CTA_PROTOINFO_TCP, __CTA_PROTOINFO_MAX }; #define CTA_PROTOINFO_MAX (__CTA_PROTOINFO_MAX - 1) +enum ctattr_protoinfo_tcp { + CTA_PROTOINFO_TCP_UNSPEC, + CTA_PROTOINFO_TCP_STATE, + __CTA_PROTOINFO_TCP_MAX +}; +#define CTA_PROTOINFO_TCP_MAX (__CTA_PROTOINFO_TCP_MAX - 1) + enum ctattr_counters { CTA_COUNTERS_UNSPEC, - CTA_COUNTERS_PACKETS, - CTA_COUNTERS_BYTES, + CTA_COUNTERS_PACKETS, /* old 64bit counters */ + CTA_COUNTERS_BYTES, /* old 64bit counters */ + CTA_COUNTERS32_PACKETS, + CTA_COUNTERS32_BYTES, __CTA_COUNTERS_MAX }; #define CTA_COUNTERS_MAX (__CTA_COUNTERS_MAX - 1) diff --git a/include/linux/netfilter_ipv4/ip_conntrack.h b/include/linux/netfilter_ipv4/ip_conntrack.h index 4ced3873681..d078bb91d9e 100644 --- a/include/linux/netfilter_ipv4/ip_conntrack.h +++ b/include/linux/netfilter_ipv4/ip_conntrack.h @@ -117,6 +117,10 @@ enum ip_conntrack_events /* NAT info */ IPCT_NATINFO_BIT = 10, IPCT_NATINFO = (1 << IPCT_NATINFO_BIT), + + /* Counter highest bit has been set */ + IPCT_COUNTER_FILLING_BIT = 11, + IPCT_COUNTER_FILLING = (1 << IPCT_COUNTER_FILLING_BIT), }; enum ip_conntrack_expect_events { @@ -192,8 +196,8 @@ do { \ struct ip_conntrack_counter { - u_int64_t packets; - u_int64_t bytes; + u_int32_t packets; + u_int32_t bytes; }; struct ip_conntrack_helper; diff --git a/include/linux/netfilter_ipv4/ip_conntrack_protocol.h b/include/linux/netfilter_ipv4/ip_conntrack_protocol.h index b6b99be8632..2c76b879e3d 100644 --- a/include/linux/netfilter_ipv4/ip_conntrack_protocol.h +++ b/include/linux/netfilter_ipv4/ip_conntrack_protocol.h @@ -52,6 +52,9 @@ struct ip_conntrack_protocol int (*to_nfattr)(struct sk_buff *skb, struct nfattr *nfa, const struct ip_conntrack *ct); + /* convert nfnetlink attributes to protoinfo */ + int (*from_nfattr)(struct nfattr *tb[], struct ip_conntrack *ct); + int (*tuple_to_nfattr)(struct sk_buff *skb, const struct ip_conntrack_tuple *t); int (*nfattr_to_tuple)(struct nfattr *tb[], diff --git a/include/linux/netfilter_ipv4/ip_conntrack_tuple.h b/include/linux/netfilter_ipv4/ip_conntrack_tuple.h index 20e43f018b7..3232db11a4e 100644 --- a/include/linux/netfilter_ipv4/ip_conntrack_tuple.h +++ b/include/linux/netfilter_ipv4/ip_conntrack_tuple.h @@ -1,6 +1,8 @@ #ifndef _IP_CONNTRACK_TUPLE_H #define _IP_CONNTRACK_TUPLE_H +#include <linux/types.h> + /* A `tuple' is a structure containing the information to uniquely identify a connection. ie. if two packets have the same tuple, they are in the same connection; if not, they are not. diff --git a/include/linux/netfilter_ipv4/ip_nat.h b/include/linux/netfilter_ipv4/ip_nat.h index e201ec6e990..41a107de17c 100644 --- a/include/linux/netfilter_ipv4/ip_nat.h +++ b/include/linux/netfilter_ipv4/ip_nat.h @@ -58,10 +58,6 @@ extern rwlock_t ip_nat_lock; struct ip_nat_info { struct list_head bysource; - - /* Helper (NULL if none). */ - struct ip_nat_helper *helper; - struct ip_nat_seq seq[IP_CT_DIR_MAX]; }; diff --git a/include/linux/netpoll.h b/include/linux/netpoll.h index 5ade54a78db..ca5a8733000 100644 --- a/include/linux/netpoll.h +++ b/include/linux/netpoll.h @@ -86,7 +86,7 @@ static inline void netpoll_poll_unlock(void *have) #else #define netpoll_rx(a) 0 -#define netpoll_poll_lock(a) 0 +#define netpoll_poll_lock(a) NULL #define netpoll_poll_unlock(a) #endif diff --git a/include/net/inet_timewait_sock.h b/include/net/inet_timewait_sock.h index 4ade56ef3a4..28f7b210350 100644 --- a/include/net/inet_timewait_sock.h +++ b/include/net/inet_timewait_sock.h @@ -19,6 +19,7 @@ #include <linux/ip.h> #include <linux/list.h> +#include <linux/module.h> #include <linux/timer.h> #include <linux/types.h> #include <linux/workqueue.h> @@ -193,11 +194,13 @@ static inline u32 inet_rcv_saddr(const struct sock *sk) static inline void inet_twsk_put(struct inet_timewait_sock *tw) { if (atomic_dec_and_test(&tw->tw_refcnt)) { + struct module *owner = tw->tw_prot->owner; #ifdef SOCK_REFCNT_DEBUG printk(KERN_DEBUG "%s timewait_sock %p released\n", tw->tw_prot->name, tw); #endif kmem_cache_free(tw->tw_prot->twsk_slab, tw); + module_put(owner); } } diff --git a/net/bridge/br_if.c b/net/bridge/br_if.c index 91bb895375f..defcf6a8607 100644 --- a/net/bridge/br_if.c +++ b/net/bridge/br_if.c @@ -79,7 +79,6 @@ static void destroy_nbp(struct net_bridge_port *p) { struct net_device *dev = p->dev; - dev->br_port = NULL; p->br = NULL; p->dev = NULL; dev_put(dev); @@ -100,6 +99,7 @@ static void del_nbp(struct net_bridge_port *p) struct net_bridge *br = p->br; struct net_device *dev = p->dev; + dev->br_port = NULL; dev_set_promiscuity(dev, -1); spin_lock_bh(&br->lock); diff --git a/net/dccp/ccid.h b/net/dccp/ccid.h index 21e55142dcd..c37eeeaf5c6 100644 --- a/net/dccp/ccid.h +++ b/net/dccp/ccid.h @@ -110,14 +110,14 @@ static inline int ccid_hc_tx_init(struct ccid *ccid, struct sock *sk) static inline void ccid_hc_rx_exit(struct ccid *ccid, struct sock *sk) { - if (ccid->ccid_hc_rx_exit != NULL && + if (ccid != NULL && ccid->ccid_hc_rx_exit != NULL && dccp_sk(sk)->dccps_hc_rx_ccid_private != NULL) ccid->ccid_hc_rx_exit(sk); } static inline void ccid_hc_tx_exit(struct ccid *ccid, struct sock *sk) { - if (ccid->ccid_hc_tx_exit != NULL && + if (ccid != NULL && ccid->ccid_hc_tx_exit != NULL && dccp_sk(sk)->dccps_hc_tx_ccid_private != NULL) ccid->ccid_hc_tx_exit(sk); } diff --git a/net/dccp/input.c b/net/dccp/input.c index 1b6b2cb1237..3454d594190 100644 --- a/net/dccp/input.c +++ b/net/dccp/input.c @@ -375,6 +375,9 @@ static int dccp_rcv_respond_partopen_state_process(struct sock *sk, case DCCP_PKT_RESET: inet_csk_clear_xmit_timer(sk, ICSK_TIME_DACK); break; + case DCCP_PKT_DATA: + if (sk->sk_state == DCCP_RESPOND) + break; case DCCP_PKT_DATAACK: case DCCP_PKT_ACK: /* @@ -393,7 +396,8 @@ static int dccp_rcv_respond_partopen_state_process(struct sock *sk, dccp_sk(sk)->dccps_osr = DCCP_SKB_CB(skb)->dccpd_seq; dccp_set_state(sk, DCCP_OPEN); - if (dh->dccph_type == DCCP_PKT_DATAACK) { + if (dh->dccph_type == DCCP_PKT_DATAACK || + dh->dccph_type == DCCP_PKT_DATA) { dccp_rcv_established(sk, skb, dh, len); queued = 1; /* packet was queued (by dccp_rcv_established) */ diff --git a/net/ipv4/esp4.c b/net/ipv4/esp4.c index 1b5a09d1b90..1b18ce66e7b 100644 --- a/net/ipv4/esp4.c +++ b/net/ipv4/esp4.c @@ -5,6 +5,7 @@ #include <net/esp.h> #include <asm/scatterlist.h> #include <linux/crypto.h> +#include <linux/kernel.h> #include <linux/pfkeyv2.h> #include <linux/random.h> #include <net/icmp.h> @@ -42,10 +43,10 @@ static int esp_output(struct xfrm_state *x, struct sk_buff *skb) esp = x->data; alen = esp->auth.icv_trunc_len; tfm = esp->conf.tfm; - blksize = (crypto_tfm_alg_blocksize(tfm) + 3) & ~3; - clen = (clen + 2 + blksize-1)&~(blksize-1); + blksize = ALIGN(crypto_tfm_alg_blocksize(tfm), 4); + clen = ALIGN(clen + 2, blksize); if (esp->conf.padlen) - clen = (clen + esp->conf.padlen-1)&~(esp->conf.padlen-1); + clen = ALIGN(clen, esp->conf.padlen); if ((nfrags = skb_cow_data(skb, clen-skb->len+alen, &trailer)) < 0) goto error; @@ -143,7 +144,7 @@ static int esp_input(struct xfrm_state *x, struct xfrm_decap_state *decap, struc struct ip_esp_hdr *esph; struct esp_data *esp = x->data; struct sk_buff *trailer; - int blksize = crypto_tfm_alg_blocksize(esp->conf.tfm); + int blksize = ALIGN(crypto_tfm_alg_blocksize(esp->conf.tfm), 4); int alen = esp->auth.icv_trunc_len; int elen = skb->len - sizeof(struct ip_esp_hdr) - esp->conf.ivlen - alen; int nfrags; @@ -304,16 +305,16 @@ static int esp_post_input(struct xfrm_state *x, struct xfrm_decap_state *decap, static u32 esp4_get_max_size(struct xfrm_state *x, int mtu) { struct esp_data *esp = x->data; - u32 blksize = crypto_tfm_alg_blocksize(esp->conf.tfm); + u32 blksize = ALIGN(crypto_tfm_alg_blocksize(esp->conf.tfm), 4); if (x->props.mode) { - mtu = (mtu + 2 + blksize-1)&~(blksize-1); + mtu = ALIGN(mtu + 2, blksize); } else { /* The worst case. */ - mtu += 2 + blksize; + mtu = ALIGN(mtu + 2, 4) + blksize - 4; } if (esp->conf.padlen) - mtu = (mtu + esp->conf.padlen-1)&~(esp->conf.padlen-1); + mtu = ALIGN(mtu, esp->conf.padlen); return mtu + x->props.header_len + esp->auth.icv_trunc_len; } diff --git a/net/ipv4/inet_timewait_sock.c b/net/ipv4/inet_timewait_sock.c index f9076ef3a1a..a010e9a6881 100644 --- a/net/ipv4/inet_timewait_sock.c +++ b/net/ipv4/inet_timewait_sock.c @@ -111,6 +111,7 @@ struct inet_timewait_sock *inet_twsk_alloc(const struct sock *sk, const int stat tw->tw_prot = sk->sk_prot_creator; atomic_set(&tw->tw_refcnt, 1); inet_twsk_dead_node_init(tw); + __module_get(tw->tw_prot->owner); } return tw; diff --git a/net/ipv4/netfilter/Kconfig b/net/ipv4/netfilter/Kconfig index a7659728e7a..7d917e4ce1d 100644 --- a/net/ipv4/netfilter/Kconfig +++ b/net/ipv4/netfilter/Kconfig @@ -139,6 +139,7 @@ config IP_NF_AMANDA config IP_NF_PPTP tristate 'PPTP protocol support' + depends on IP_NF_CONNTRACK help This module adds support for PPTP (Point to Point Tunnelling Protocol, RFC2637) connection tracking and NAT. @@ -498,9 +499,14 @@ config IP_NF_TARGET_LOG To compile it as a module, choose M here. If unsure, say N. config IP_NF_TARGET_ULOG - tristate "ULOG target support" + tristate "ULOG target support (OBSOLETE)" depends on IP_NF_IPTABLES ---help--- + + This option enables the old IPv4-only "ipt_ULOG" implementation + which has been obsoleted by the new "nfnetlink_log" code (see + CONFIG_NETFILTER_NETLINK_LOG). + This option adds a `ULOG' target, which allows you to create rules in any iptables table. The packet is passed to a userspace logging daemon using netlink multicast sockets; unlike the LOG target diff --git a/net/ipv4/netfilter/ip_conntrack_core.c b/net/ipv4/netfilter/ip_conntrack_core.c index ea65dd3e517..07a80b56e8d 100644 --- a/net/ipv4/netfilter/ip_conntrack_core.c +++ b/net/ipv4/netfilter/ip_conntrack_core.c @@ -1119,7 +1119,7 @@ void __ip_ct_refresh_acct(struct ip_conntrack *ct, unsigned long extra_jiffies, int do_acct) { - int do_event = 0; + int event = 0; IP_NF_ASSERT(ct->timeout.data == (unsigned long)ct); IP_NF_ASSERT(skb); @@ -1129,13 +1129,13 @@ void __ip_ct_refresh_acct(struct ip_conntrack *ct, /* If not in hash table, timer will not be active yet */ if (!is_confirmed(ct)) { ct->timeout.expires = extra_jiffies; - do_event = 1; + event = IPCT_REFRESH; } else { /* Need del_timer for race avoidance (may already be dying). */ if (del_timer(&ct->timeout)) { ct->timeout.expires = jiffies + extra_jiffies; add_timer(&ct->timeout); - do_event = 1; + event = IPCT_REFRESH; } } @@ -1144,14 +1144,17 @@ void __ip_ct_refresh_acct(struct ip_conntrack *ct, ct->counters[CTINFO2DIR(ctinfo)].packets++; ct->counters[CTINFO2DIR(ctinfo)].bytes += ntohs(skb->nh.iph->tot_len); + if ((ct->counters[CTINFO2DIR(ctinfo)].packets & 0x80000000) + || (ct->counters[CTINFO2DIR(ctinfo)].bytes & 0x80000000)) + event |= IPCT_COUNTER_FILLING; } #endif write_unlock_bh(&ip_conntrack_lock); /* must be unlocked when calling event cache */ - if (do_event) - ip_conntrack_event_cache(IPCT_REFRESH, skb); + if (event) + ip_conntrack_event_cache(event, skb); } #if defined(CONFIG_IP_NF_CONNTRACK_NETLINK) || \ diff --git a/net/ipv4/netfilter/ip_conntrack_netlink.c b/net/ipv4/netfilter/ip_conntrack_netlink.c index b08a432efcf..166e6069f12 100644 --- a/net/ipv4/netfilter/ip_conntrack_netlink.c +++ b/net/ipv4/netfilter/ip_conntrack_netlink.c @@ -177,11 +177,11 @@ ctnetlink_dump_counters(struct sk_buff *skb, const struct ip_conntrack *ct, struct nfattr *nest_count = NFA_NEST(skb, type); u_int64_t tmp; - tmp = cpu_to_be64(ct->counters[dir].packets); - NFA_PUT(skb, CTA_COUNTERS_PACKETS, sizeof(u_int64_t), &tmp); + tmp = htonl(ct->counters[dir].packets); + NFA_PUT(skb, CTA_COUNTERS32_PACKETS, sizeof(u_int32_t), &tmp); - tmp = cpu_to_be64(ct->counters[dir].bytes); - NFA_PUT(skb, CTA_COUNTERS_BYTES, sizeof(u_int64_t), &tmp); + tmp = htonl(ct->counters[dir].bytes); + NFA_PUT(skb, CTA_COUNTERS32_BYTES, sizeof(u_int32_t), &tmp); NFA_NEST_END(skb, nest_count); @@ -833,7 +833,8 @@ out: static inline int ctnetlink_change_status(struct ip_conntrack *ct, struct nfattr *cda[]) { - unsigned long d, status = *(u_int32_t *)NFA_DATA(cda[CTA_STATUS-1]); + unsigned long d; + unsigned status = ntohl(*(u_int32_t *)NFA_DATA(cda[CTA_STATUS-1])); d = ct->status ^ status; if (d & (IPS_EXPECTED|IPS_CONFIRMED|IPS_DYING)) @@ -948,6 +949,31 @@ ctnetlink_change_timeout(struct ip_conntrack *ct, struct nfattr *cda[]) return 0; } +static inline int +ctnetlink_change_protoinfo(struct ip_conntrack *ct, struct nfattr *cda[]) +{ + struct nfattr *tb[CTA_PROTOINFO_MAX], *attr = cda[CTA_PROTOINFO-1]; + struct ip_conntrack_protocol *proto; + u_int16_t npt = ct->tuplehash[IP_CT_DIR_ORIGINAL].tuple.dst.protonum; + int err = 0; + + if (nfattr_parse_nested(tb, CTA_PROTOINFO_MAX, attr) < 0) + goto nfattr_failure; + + proto = ip_conntrack_proto_find_get(npt); + if (!proto) + return -EINVAL; + + if (proto->from_nfattr) + err = proto->from_nfattr(tb, ct); + ip_conntrack_proto_put(proto); + + return err; + +nfattr_failure: + return -ENOMEM; +} + static int ctnetlink_change_conntrack(struct ip_conntrack *ct, struct nfattr *cda[]) { @@ -973,6 +999,12 @@ ctnetlink_change_conntrack(struct ip_conntrack *ct, struct nfattr *cda[]) return err; } + if (cda[CTA_PROTOINFO-1]) { + err = ctnetlink_change_protoinfo(ct, cda); + if (err < 0) + return err; + } + DEBUGP("all done\n"); return 0; } @@ -1002,6 +1034,12 @@ ctnetlink_create_conntrack(struct nfattr *cda[], if (err < 0) goto err; + if (cda[CTA_PROTOINFO-1]) { + err = ctnetlink_change_protoinfo(ct, cda); + if (err < 0) + return err; + } + ct->helper = ip_conntrack_helper_find_get(rtuple); add_timer(&ct->timeout); diff --git a/net/ipv4/netfilter/ip_conntrack_proto_icmp.c b/net/ipv4/netfilter/ip_conntrack_proto_icmp.c index 838d1d69b36..98f0015dd25 100644 --- a/net/ipv4/netfilter/ip_conntrack_proto_icmp.c +++ b/net/ipv4/netfilter/ip_conntrack_proto_icmp.c @@ -296,8 +296,7 @@ static int icmp_nfattr_to_tuple(struct nfattr *tb[], struct ip_conntrack_tuple *tuple) { if (!tb[CTA_PROTO_ICMP_TYPE-1] - || !tb[CTA_PROTO_ICMP_CODE-1] - || !tb[CTA_PROTO_ICMP_ID-1]) + || !tb[CTA_PROTO_ICMP_CODE-1]) return -1; tuple->dst.u.icmp.type = diff --git a/net/ipv4/netfilter/ip_conntrack_proto_tcp.c b/net/ipv4/netfilter/ip_conntrack_proto_tcp.c index 121760d6cc5..d6701cafbcc 100644 --- a/net/ipv4/netfilter/ip_conntrack_proto_tcp.c +++ b/net/ipv4/netfilter/ip_conntrack_proto_tcp.c @@ -341,17 +341,43 @@ static int tcp_print_conntrack(struct seq_file *s, static int tcp_to_nfattr(struct sk_buff *skb, struct nfattr *nfa, const struct ip_conntrack *ct) { + struct nfattr *nest_parms = NFA_NEST(skb, CTA_PROTOINFO_TCP); + read_lock_bh(&tcp_lock); NFA_PUT(skb, CTA_PROTOINFO_TCP_STATE, sizeof(u_int8_t), &ct->proto.tcp.state); read_unlock_bh(&tcp_lock); + NFA_NEST_END(skb, nest_parms); + return 0; nfattr_failure: read_unlock_bh(&tcp_lock); return -1; } + +static int nfattr_to_tcp(struct nfattr *cda[], struct ip_conntrack *ct) +{ + struct nfattr *attr = cda[CTA_PROTOINFO_TCP-1]; + struct nfattr *tb[CTA_PROTOINFO_TCP_MAX]; + + if (nfattr_parse_nested(tb, CTA_PROTOINFO_TCP_MAX, attr) < 0) + goto nfattr_failure; + + if (!tb[CTA_PROTOINFO_TCP_STATE-1]) + return -EINVAL; + + write_lock_bh(&tcp_lock); + ct->proto.tcp.state = + *(u_int8_t *)NFA_DATA(tb[CTA_PROTOINFO_TCP_STATE-1]); + write_unlock_bh(&tcp_lock); + + return 0; + +nfattr_failure: + return -1; +} #endif static unsigned int get_conntrack_index(const struct tcphdr *tcph) @@ -1123,6 +1149,7 @@ struct ip_conntrack_protocol ip_conntrack_protocol_tcp = #if defined(CONFIG_IP_NF_CONNTRACK_NETLINK) || \ defined(CONFIG_IP_NF_CONNTRACK_NETLINK_MODULE) .to_nfattr = tcp_to_nfattr, + .from_nfattr = nfattr_to_tcp, .tuple_to_nfattr = ip_ct_port_tuple_to_nfattr, .nfattr_to_tuple = ip_ct_port_nfattr_to_tuple, #endif diff --git a/net/ipv4/tcp_output.c b/net/ipv4/tcp_output.c index 8225e425725..f37a50e55b6 100644 --- a/net/ipv4/tcp_output.c +++ b/net/ipv4/tcp_output.c @@ -435,7 +435,14 @@ int tcp_fragment(struct sock *sk, struct sk_buff *skb, u32 len, unsigned int mss int nsize, old_factor; u16 flags; - BUG_ON(len >= skb->len); + if (unlikely(len >= skb->len)) { + printk(KERN_DEBUG "TCP: seg_size=%u, mss=%u, seq=%u, " + "end_seq=%u, skb->len=%u.\n", len, mss_now, + TCP_SKB_CB(skb)->seq, TCP_SKB_CB(skb)->end_seq, + skb->len); + WARN_ON(1); + return 0; + } nsize = skb_headlen(skb) - len; if (nsize < 0) diff --git a/net/ipv6/esp6.c b/net/ipv6/esp6.c index 9b27460f0cc..40d9a1935ab 100644 --- a/net/ipv6/esp6.c +++ b/net/ipv6/esp6.c @@ -31,6 +31,7 @@ #include <net/esp.h> #include <asm/scatterlist.h> #include <linux/crypto.h> +#include <linux/kernel.h> #include <linux/pfkeyv2.h> #include <linux/random.h> #include <net/icmp.h> @@ -66,10 +67,10 @@ static int esp6_output(struct xfrm_state *x, struct sk_buff *skb) alen = esp->auth.icv_trunc_len; tfm = esp->conf.tfm; - blksize = (crypto_tfm_alg_blocksize(tfm) + 3) & ~3; - clen = (clen + 2 + blksize-1)&~(blksize-1); + blksize = ALIGN(crypto_tfm_alg_blocksize(tfm), 4); + clen = ALIGN(clen + 2, blksize); if (esp->conf.padlen) - clen = (clen + esp->conf.padlen-1)&~(esp->conf.padlen-1); + clen = ALIGN(clen, esp->conf.padlen); if ((nfrags = skb_cow_data(skb, clen-skb->len+alen, &trailer)) < 0) { goto error; @@ -133,7 +134,7 @@ static int esp6_input(struct xfrm_state *x, struct xfrm_decap_state *decap, stru struct ipv6_esp_hdr *esph; struct esp_data *esp = x->data; struct sk_buff *trailer; - int blksize = crypto_tfm_alg_blocksize(esp->conf.tfm); + int blksize = ALIGN(crypto_tfm_alg_blocksize(esp->conf.tfm), 4); int alen = esp->auth.icv_trunc_len; int elen = skb->len - sizeof(struct ipv6_esp_hdr) - esp->conf.ivlen - alen; @@ -235,16 +236,17 @@ out_nofree: static u32 esp6_get_max_size(struct xfrm_state *x, int mtu) { struct esp_data *esp = x->data; - u32 blksize = crypto_tfm_alg_blocksize(esp->conf.tfm); + u32 blksize = ALIGN(crypto_tfm_alg_blocksize(esp->conf.tfm), 4); if (x->props.mode) { - mtu = (mtu + 2 + blksize-1)&~(blksize-1); + mtu = ALIGN(mtu + 2, blksize); } else { /* The worst case. */ - mtu += 2 + blksize; + u32 padsize = ((blksize - 1) & 7) + 1; + mtu = ALIGN(mtu + 2, padsize) + blksize - padsize; } if (esp->conf.padlen) - mtu = (mtu + esp->conf.padlen-1)&~(esp->conf.padlen-1); + mtu = ALIGN(mtu, esp->conf.padlen); return mtu + x->props.header_len + esp->auth.icv_full_len; } diff --git a/net/netfilter/nfnetlink.c b/net/netfilter/nfnetlink.c index 1caaca06f69..4bc27a6334c 100644 --- a/net/netfilter/nfnetlink.c +++ b/net/netfilter/nfnetlink.c @@ -133,7 +133,7 @@ int nfattr_parse(struct nfattr *tb[], int maxattr, struct nfattr *nfa, int len) memset(tb, 0, sizeof(struct nfattr *) * maxattr); while (NFA_OK(nfa, len)) { - unsigned flavor = nfa->nfa_type; + unsigned flavor = NFA_TYPE(nfa); if (flavor && flavor <= maxattr) tb[flavor-1] = nfa; nfa = NFA_NEXT(nfa, len); @@ -177,7 +177,7 @@ nfnetlink_check_attributes(struct nfnetlink_subsystem *subsys, int attrlen = nlh->nlmsg_len - NLMSG_ALIGN(min_len); while (NFA_OK(attr, attrlen)) { - unsigned flavor = attr->nfa_type; + unsigned flavor = NFA_TYPE(attr); if (flavor) { if (flavor > attr_count) return -EINVAL; |