diff options
Diffstat (limited to 'arch')
25 files changed, 305 insertions, 207 deletions
diff --git a/arch/ia64/ia32/sys_ia32.c b/arch/ia64/ia32/sys_ia32.c index 247a21c64aea..c1e20d65dd6c 100644 --- a/arch/ia64/ia32/sys_ia32.c +++ b/arch/ia64/ia32/sys_ia32.c @@ -2427,7 +2427,7 @@ sys32_epoll_wait(int epfd, struct epoll_event32 __user * events, int maxevents, { struct epoll_event *events64 = NULL; mm_segment_t old_fs = get_fs(); - int error, numevents, size; + int numevents, size; int evt_idx; int do_free_pages = 0; diff --git a/arch/ia64/kernel/mca.c b/arch/ia64/kernel/mca.c index 2c75741dcc66..736e328b5e61 100644 --- a/arch/ia64/kernel/mca.c +++ b/arch/ia64/kernel/mca.c @@ -1103,8 +1103,6 @@ ia64_mca_cpe_int_caller(int cpe_irq, void *arg, struct pt_regs *ptregs) return IRQ_HANDLED; } -#endif /* CONFIG_ACPI */ - /* * ia64_mca_cpe_poll * @@ -1122,6 +1120,8 @@ ia64_mca_cpe_poll (unsigned long dummy) platform_send_ipi(first_cpu(cpu_online_map), IA64_CPEP_VECTOR, IA64_IPI_DM_INT, 0); } +#endif /* CONFIG_ACPI */ + /* * C portion of the OS INIT handler * diff --git a/arch/m68knommu/kernel/process.c b/arch/m68knommu/kernel/process.c index 2b6c9d32b7a6..c4a33f265dc0 100644 --- a/arch/m68knommu/kernel/process.c +++ b/arch/m68knommu/kernel/process.c @@ -45,11 +45,13 @@ asmlinkage void ret_from_fork(void); */ void default_idle(void) { - while(1) { - if (need_resched()) - __asm__("stop #0x2000" : : : "cc"); - schedule(); + local_irq_disable(); + while (!need_resched()) { + /* This stop will re-enable interrupts */ + __asm__("stop #0x2000" : : : "cc"); + local_irq_disable(); } + local_irq_enable(); } void (*idle)(void) = default_idle; @@ -63,7 +65,12 @@ void (*idle)(void) = default_idle; void cpu_idle(void) { /* endless idle loop with no priority at all */ - idle(); + while (1) { + idle(); + preempt_enable_no_resched(); + schedule(); + preempt_disable(); + } } void machine_restart(char * __unused) diff --git a/arch/ppc/kernel/cputable.c b/arch/ppc/kernel/cputable.c index 8aa5e8c69009..d44b7dc5390a 100644 --- a/arch/ppc/kernel/cputable.c +++ b/arch/ppc/kernel/cputable.c @@ -838,6 +838,17 @@ struct cpu_spec cpu_specs[] = { .icache_bsize = 32, .dcache_bsize = 32, }, + { /* 405EP */ + .pvr_mask = 0xffff0000, + .pvr_value = 0x51210000, + .cpu_name = "405EP", + .cpu_features = CPU_FTR_SPLIT_ID_CACHE | + CPU_FTR_USE_TB, + .cpu_user_features = PPC_FEATURE_32 | + PPC_FEATURE_HAS_MMU | PPC_FEATURE_HAS_4xxMAC, + .icache_bsize = 32, + .dcache_bsize = 32, + }, #endif /* CONFIG_40x */ #ifdef CONFIG_44x diff --git a/arch/ppc/kernel/misc.S b/arch/ppc/kernel/misc.S index e4f1615ec13f..7329ef177a18 100644 --- a/arch/ppc/kernel/misc.S +++ b/arch/ppc/kernel/misc.S @@ -619,7 +619,7 @@ _GLOBAL(flush_instruction_cache) _GLOBAL(flush_icache_range) BEGIN_FTR_SECTION blr /* for 601, do nothing */ -END_FTR_SECTION_IFSET(PPC_FEATURE_UNIFIED_CACHE) +END_FTR_SECTION_IFCLR(CPU_FTR_SPLIT_ID_CACHE) li r5,L1_CACHE_LINE_SIZE-1 andc r3,r3,r5 subf r4,r3,r4 @@ -736,7 +736,7 @@ _GLOBAL(flush_dcache_all) _GLOBAL(__flush_dcache_icache) BEGIN_FTR_SECTION blr /* for 601, do nothing */ -END_FTR_SECTION_IFSET(PPC_FEATURE_UNIFIED_CACHE) +END_FTR_SECTION_IFCLR(CPU_FTR_SPLIT_ID_CACHE) rlwinm r3,r3,0,0,19 /* Get page base address */ li r4,4096/L1_CACHE_LINE_SIZE /* Number of lines in a page */ mtctr r4 @@ -764,7 +764,7 @@ END_FTR_SECTION_IFSET(PPC_FEATURE_UNIFIED_CACHE) _GLOBAL(__flush_dcache_icache_phys) BEGIN_FTR_SECTION blr /* for 601, do nothing */ -END_FTR_SECTION_IFSET(PPC_FEATURE_UNIFIED_CACHE) +END_FTR_SECTION_IFCLR(CPU_FTR_SPLIT_ID_CACHE) mfmsr r10 rlwinm r0,r10,0,28,26 /* clear DR */ mtmsr r0 diff --git a/arch/ppc64/kernel/entry.S b/arch/ppc64/kernel/entry.S index d3604056e1a9..b61572eb2a71 100644 --- a/arch/ppc64/kernel/entry.S +++ b/arch/ppc64/kernel/entry.S @@ -436,15 +436,6 @@ END_FTR_SECTION_IFSET(CPU_FTR_ALTIVEC) REST_8GPRS(14, r1) REST_10GPRS(22, r1) -#ifdef CONFIG_PPC_ISERIES - clrrdi r7,r1,THREAD_SHIFT /* get current_thread_info() */ - ld r7,TI_FLAGS(r7) /* Get run light flag */ - mfspr r9,CTRLF - srdi r7,r7,TIF_RUN_LIGHT - insrdi r9,r7,1,63 /* Insert run light into CTRL */ - mtspr CTRLT,r9 -#endif - /* convert old thread to its task_struct for return value */ addi r3,r3,-THREAD ld r7,_NIP(r1) /* Return to _switch caller in new task */ diff --git a/arch/ppc64/kernel/head.S b/arch/ppc64/kernel/head.S index 92a744c31ab1..346dbf606b5d 100644 --- a/arch/ppc64/kernel/head.S +++ b/arch/ppc64/kernel/head.S @@ -626,10 +626,10 @@ system_reset_iSeries: lhz r24,PACAPACAINDEX(r13) /* Get processor # */ cmpwi 0,r24,0 /* Are we processor 0? */ beq .__start_initialization_iSeries /* Start up the first processor */ - mfspr r4,CTRLF - li r5,RUNLATCH /* Turn off the run light */ + mfspr r4,SPRN_CTRLF + li r5,CTRL_RUNLATCH /* Turn off the run light */ andc r4,r4,r5 - mtspr CTRLT,r4 + mtspr SPRN_CTRLT,r4 1: HMT_LOW @@ -2082,9 +2082,9 @@ _GLOBAL(hmt_start_secondary) mfspr r4, HID0 ori r4, r4, 0x1 mtspr HID0, r4 - mfspr r4, CTRLF + mfspr r4, SPRN_CTRLF oris r4, r4, 0x40 - mtspr CTRLT, r4 + mtspr SPRN_CTRLT, r4 blr #endif diff --git a/arch/ppc64/kernel/iSeries_setup.c b/arch/ppc64/kernel/iSeries_setup.c index da20120f2261..6d06eb550a3f 100644 --- a/arch/ppc64/kernel/iSeries_setup.c +++ b/arch/ppc64/kernel/iSeries_setup.c @@ -852,6 +852,28 @@ static int __init iSeries_src_init(void) late_initcall(iSeries_src_init); +static int set_spread_lpevents(char *str) +{ + unsigned long i; + unsigned long val = simple_strtoul(str, NULL, 0); + + /* + * The parameter is the number of processors to share in processing + * lp events. + */ + if (( val > 0) && (val <= NR_CPUS)) { + for (i = 1; i < val; ++i) + paca[i].lpqueue_ptr = paca[0].lpqueue_ptr; + + printk("lpevent processing spread over %ld processors\n", val); + } else { + printk("invalid spread_lpevents %ld\n", val); + } + + return 1; +} +__setup("spread_lpevents=", set_spread_lpevents); + void __init iSeries_early_setup(void) { iSeries_fixup_klimit(); diff --git a/arch/ppc64/kernel/idle.c b/arch/ppc64/kernel/idle.c index 6abc621d3ba0..f24ce2b87200 100644 --- a/arch/ppc64/kernel/idle.c +++ b/arch/ppc64/kernel/idle.c @@ -75,13 +75,9 @@ static int iSeries_idle(void) { struct paca_struct *lpaca; long oldval; - unsigned long CTRL; /* ensure iSeries run light will be out when idle */ - clear_thread_flag(TIF_RUN_LIGHT); - CTRL = mfspr(CTRLF); - CTRL &= ~RUNLATCH; - mtspr(CTRLT, CTRL); + ppc64_runlatch_off(); lpaca = get_paca(); @@ -111,7 +107,9 @@ static int iSeries_idle(void) } } + ppc64_runlatch_on(); schedule(); + ppc64_runlatch_off(); } return 0; diff --git a/arch/ppc64/kernel/process.c b/arch/ppc64/kernel/process.c index 8b0686122738..cdfecbeb331f 100644 --- a/arch/ppc64/kernel/process.c +++ b/arch/ppc64/kernel/process.c @@ -378,9 +378,6 @@ copy_thread(int nr, unsigned long clone_flags, unsigned long usp, childregs->gpr[1] = sp + sizeof(struct pt_regs); p->thread.regs = NULL; /* no user register state */ clear_ti_thread_flag(p->thread_info, TIF_32BIT); -#ifdef CONFIG_PPC_ISERIES - set_ti_thread_flag(p->thread_info, TIF_RUN_LIGHT); -#endif } else { childregs->gpr[1] = usp; p->thread.regs = childregs; diff --git a/arch/ppc64/kernel/prom_init.c b/arch/ppc64/kernel/prom_init.c index 3de950de3671..b7683abfbe6a 100644 --- a/arch/ppc64/kernel/prom_init.c +++ b/arch/ppc64/kernel/prom_init.c @@ -211,13 +211,23 @@ struct { */ #define ADDR(x) (u32) ((unsigned long)(x) - offset) +/* + * Error results ... some OF calls will return "-1" on error, some + * will return 0, some will return either. To simplify, here are + * macros to use with any ihandle or phandle return value to check if + * it is valid + */ + +#define PROM_ERROR (-1u) +#define PHANDLE_VALID(p) ((p) != 0 && (p) != PROM_ERROR) +#define IHANDLE_VALID(i) ((i) != 0 && (i) != PROM_ERROR) + + /* This is the one and *ONLY* place where we actually call open * firmware from, since we need to make sure we're running in 32b * mode when we do. We switch back to 64b mode upon return. */ -#define PROM_ERROR (-1) - static int __init call_prom(const char *service, int nargs, int nret, ...) { int i; @@ -587,14 +597,13 @@ static void __init prom_send_capabilities(void) { unsigned long offset = reloc_offset(); ihandle elfloader; - int ret; elfloader = call_prom("open", 1, 1, ADDR("/packages/elf-loader")); if (elfloader == 0) { prom_printf("couldn't open /packages/elf-loader\n"); return; } - ret = call_prom("call-method", 3, 1, ADDR("process-elf-header"), + call_prom("call-method", 3, 1, ADDR("process-elf-header"), elfloader, ADDR(&fake_elf)); call_prom("close", 1, 0, elfloader); } @@ -646,7 +655,7 @@ static unsigned long __init alloc_up(unsigned long size, unsigned long align) base = _ALIGN_UP(base + 0x100000, align)) { prom_debug(" trying: 0x%x\n\r", base); addr = (unsigned long)prom_claim(base, size, 0); - if ((int)addr != PROM_ERROR) + if (addr != PROM_ERROR) break; addr = 0; if (align == 0) @@ -708,7 +717,7 @@ static unsigned long __init alloc_down(unsigned long size, unsigned long align, for(; base > RELOC(alloc_bottom); base = _ALIGN_DOWN(base - 0x100000, align)) { prom_debug(" trying: 0x%x\n\r", base); addr = (unsigned long)prom_claim(base, size, 0); - if ((int)addr != PROM_ERROR) + if (addr != PROM_ERROR) break; addr = 0; } @@ -902,18 +911,19 @@ static void __init prom_instantiate_rtas(void) { unsigned long offset = reloc_offset(); struct prom_t *_prom = PTRRELOC(&prom); - phandle prom_rtas, rtas_node; + phandle rtas_node; + ihandle rtas_inst; u32 base, entry = 0; u32 size = 0; prom_debug("prom_instantiate_rtas: start...\n"); - prom_rtas = call_prom("finddevice", 1, 1, ADDR("/rtas")); - prom_debug("prom_rtas: %x\n", prom_rtas); - if (prom_rtas == (phandle) -1) + rtas_node = call_prom("finddevice", 1, 1, ADDR("/rtas")); + prom_debug("rtas_node: %x\n", rtas_node); + if (!PHANDLE_VALID(rtas_node)) return; - prom_getprop(prom_rtas, "rtas-size", &size, sizeof(size)); + prom_getprop(rtas_node, "rtas-size", &size, sizeof(size)); if (size == 0) return; @@ -922,14 +932,18 @@ static void __init prom_instantiate_rtas(void) prom_printf("RTAS allocation failed !\n"); return; } - prom_printf("instantiating rtas at 0x%x", base); - rtas_node = call_prom("open", 1, 1, ADDR("/rtas")); - prom_printf("..."); + rtas_inst = call_prom("open", 1, 1, ADDR("/rtas")); + if (!IHANDLE_VALID(rtas_inst)) { + prom_printf("opening rtas package failed"); + return; + } + + prom_printf("instantiating rtas at 0x%x ...", base); if (call_prom("call-method", 3, 2, ADDR("instantiate-rtas"), - rtas_node, base) != PROM_ERROR) { + rtas_inst, base) != PROM_ERROR) { entry = (long)_prom->args.rets[1]; } if (entry == 0) { @@ -940,8 +954,8 @@ static void __init prom_instantiate_rtas(void) reserve_mem(base, size); - prom_setprop(prom_rtas, "linux,rtas-base", &base, sizeof(base)); - prom_setprop(prom_rtas, "linux,rtas-entry", &entry, sizeof(entry)); + prom_setprop(rtas_node, "linux,rtas-base", &base, sizeof(base)); + prom_setprop(rtas_node, "linux,rtas-entry", &entry, sizeof(entry)); prom_debug("rtas base = 0x%x\n", base); prom_debug("rtas entry = 0x%x\n", entry); @@ -1062,7 +1076,7 @@ static void __init prom_initialize_tce_table(void) prom_printf("opening PHB %s", path); phb_node = call_prom("open", 1, 1, path); - if ( (long)phb_node <= 0) + if (phb_node == 0) prom_printf("... failed\n"); else prom_printf("... done\n"); @@ -1279,12 +1293,12 @@ static void __init prom_init_client_services(unsigned long pp) /* get a handle for the stdout device */ _prom->chosen = call_prom("finddevice", 1, 1, ADDR("/chosen")); - if ((long)_prom->chosen <= 0) + if (!PHANDLE_VALID(_prom->chosen)) prom_panic("cannot find chosen"); /* msg won't be printed :( */ /* get device tree root */ _prom->root = call_prom("finddevice", 1, 1, ADDR("/")); - if ((long)_prom->root <= 0) + if (!PHANDLE_VALID(_prom->root)) prom_panic("cannot find device tree root"); /* msg won't be printed :( */ } @@ -1356,9 +1370,8 @@ static int __init prom_find_machine_type(void) } /* Default to pSeries. We need to know if we are running LPAR */ rtas = call_prom("finddevice", 1, 1, ADDR("/rtas")); - if (rtas != (phandle) -1) { - unsigned long x; - x = prom_getproplen(rtas, "ibm,hypertas-functions"); + if (PHANDLE_VALID(rtas)) { + int x = prom_getproplen(rtas, "ibm,hypertas-functions"); if (x != PROM_ERROR) { prom_printf("Hypertas detected, assuming LPAR !\n"); return PLATFORM_PSERIES_LPAR; @@ -1426,12 +1439,13 @@ static void __init prom_check_displays(void) * leave some room at the end of the path for appending extra * arguments */ - if (call_prom("package-to-path", 3, 1, node, path, PROM_SCRATCH_SIZE-10) < 0) + if (call_prom("package-to-path", 3, 1, node, path, + PROM_SCRATCH_SIZE-10) == PROM_ERROR) continue; prom_printf("found display : %s, opening ... ", path); ih = call_prom("open", 1, 1, path); - if (ih == (ihandle)0 || ih == (ihandle)-1) { + if (ih == 0) { prom_printf("failed\n"); continue; } @@ -1514,6 +1528,12 @@ static unsigned long __init dt_find_string(char *str) return 0; } +/* + * The Open Firmware 1275 specification states properties must be 31 bytes or + * less, however not all firmwares obey this. Make it 64 bytes to be safe. + */ +#define MAX_PROPERTY_NAME 64 + static void __init scan_dt_build_strings(phandle node, unsigned long *mem_start, unsigned long *mem_end) { @@ -1527,10 +1547,12 @@ static void __init scan_dt_build_strings(phandle node, unsigned long *mem_start, /* get and store all property names */ prev_name = RELOC(""); for (;;) { - - /* 32 is max len of name including nul. */ - namep = make_room(mem_start, mem_end, 32, 1); - if (call_prom("nextprop", 3, 1, node, prev_name, namep) <= 0) { + int rc; + + /* 64 is max len of name including nul. */ + namep = make_room(mem_start, mem_end, MAX_PROPERTY_NAME, 1); + rc = call_prom("nextprop", 3, 1, node, prev_name, namep); + if (rc != 1) { /* No more nodes: unwind alloc */ *mem_start = (unsigned long)namep; break; @@ -1555,12 +1577,6 @@ static void __init scan_dt_build_strings(phandle node, unsigned long *mem_start, } } -/* - * The Open Firmware 1275 specification states properties must be 31 bytes or - * less, however not all firmwares obey this. Make it 64 bytes to be safe. - */ -#define MAX_PROPERTY_NAME 64 - static void __init scan_dt_build_struct(phandle node, unsigned long *mem_start, unsigned long *mem_end) { @@ -1607,7 +1623,10 @@ static void __init scan_dt_build_struct(phandle node, unsigned long *mem_start, prev_name = RELOC(""); sstart = (char *)RELOC(dt_string_start); for (;;) { - if (call_prom("nextprop", 3, 1, node, prev_name, pname) <= 0) + int rc; + + rc = call_prom("nextprop", 3, 1, node, prev_name, pname); + if (rc != 1) break; /* find string offset */ @@ -1623,7 +1642,7 @@ static void __init scan_dt_build_struct(phandle node, unsigned long *mem_start, l = call_prom("getproplen", 2, 1, node, pname); /* sanity checks */ - if (l < 0) + if (l == PROM_ERROR) continue; if (l > MAX_PROPERTY_LENGTH) { prom_printf("WARNING: ignoring large property "); @@ -1771,17 +1790,18 @@ static void __init fixup_device_tree(void) /* Some G5s have a missing interrupt definition, fix it up here */ u3 = call_prom("finddevice", 1, 1, ADDR("/u3@0,f8000000")); - if ((long)u3 <= 0) + if (!PHANDLE_VALID(u3)) return; i2c = call_prom("finddevice", 1, 1, ADDR("/u3@0,f8000000/i2c@f8001000")); - if ((long)i2c <= 0) + if (!PHANDLE_VALID(i2c)) return; mpic = call_prom("finddevice", 1, 1, ADDR("/u3@0,f8000000/mpic@f8040000")); - if ((long)mpic <= 0) + if (!PHANDLE_VALID(mpic)) return; /* check if proper rev of u3 */ - if (prom_getprop(u3, "device-rev", &u3_rev, sizeof(u3_rev)) <= 0) + if (prom_getprop(u3, "device-rev", &u3_rev, sizeof(u3_rev)) + == PROM_ERROR) return; if (u3_rev != 0x35) return; diff --git a/arch/ppc64/kernel/setup.c b/arch/ppc64/kernel/setup.c index 21c57f539c29..dce198d39328 100644 --- a/arch/ppc64/kernel/setup.c +++ b/arch/ppc64/kernel/setup.c @@ -103,11 +103,6 @@ extern void unflatten_device_tree(void); extern void smp_release_cpus(void); -unsigned long decr_overclock = 1; -unsigned long decr_overclock_proc0 = 1; -unsigned long decr_overclock_set = 0; -unsigned long decr_overclock_proc0_set = 0; - int have_of = 1; int boot_cpuid = 0; int boot_cpuid_phys = 0; @@ -1120,64 +1115,15 @@ void ppc64_dump_msg(unsigned int src, const char *msg) printk("[dump]%04x %s\n", src, msg); } -int set_spread_lpevents( char * str ) -{ - /* The parameter is the number of processors to share in processing lp events */ - unsigned long i; - unsigned long val = simple_strtoul( str, NULL, 0 ); - if ( ( val > 0 ) && ( val <= NR_CPUS ) ) { - for ( i=1; i<val; ++i ) - paca[i].lpqueue_ptr = paca[0].lpqueue_ptr; - printk("lpevent processing spread over %ld processors\n", val); - } - else - printk("invalid spreaqd_lpevents %ld\n", val); - return 1; -} - /* This should only be called on processor 0 during calibrate decr */ void setup_default_decr(void) { struct paca_struct *lpaca = get_paca(); - if ( decr_overclock_set && !decr_overclock_proc0_set ) - decr_overclock_proc0 = decr_overclock; - - lpaca->default_decr = tb_ticks_per_jiffy / decr_overclock_proc0; + lpaca->default_decr = tb_ticks_per_jiffy; lpaca->next_jiffy_update_tb = get_tb() + tb_ticks_per_jiffy; } -int set_decr_overclock_proc0( char * str ) -{ - unsigned long val = simple_strtoul( str, NULL, 0 ); - if ( ( val >= 1 ) && ( val <= 48 ) ) { - decr_overclock_proc0_set = 1; - decr_overclock_proc0 = val; - printk("proc 0 decrementer overclock factor of %ld\n", val); - } - else - printk("invalid proc 0 decrementer overclock factor of %ld\n", val); - return 1; -} - -int set_decr_overclock( char * str ) -{ - unsigned long val = simple_strtoul( str, NULL, 0 ); - if ( ( val >= 1 ) && ( val <= 48 ) ) { - decr_overclock_set = 1; - decr_overclock = val; - printk("decrementer overclock factor of %ld\n", val); - } - else - printk("invalid decrementer overclock factor of %ld\n", val); - return 1; - -} - -__setup("spread_lpevents=", set_spread_lpevents ); -__setup("decr_overclock_proc0=", set_decr_overclock_proc0 ); -__setup("decr_overclock=", set_decr_overclock ); - #ifndef CONFIG_PPC_ISERIES /* * This function can be used by platforms to "find" legacy serial ports. diff --git a/arch/ppc64/kernel/smp.c b/arch/ppc64/kernel/smp.c index 3b906cd94037..9ef5d36d6b25 100644 --- a/arch/ppc64/kernel/smp.c +++ b/arch/ppc64/kernel/smp.c @@ -334,7 +334,6 @@ void smp_call_function_interrupt(void) } } -extern unsigned long decr_overclock; extern struct gettimeofday_struct do_gtod; struct thread_info *current_set[NR_CPUS]; @@ -491,7 +490,7 @@ int __devinit __cpu_up(unsigned int cpu) if (smp_ops->cpu_bootable && !smp_ops->cpu_bootable(cpu)) return -EINVAL; - paca[cpu].default_decr = tb_ticks_per_jiffy / decr_overclock; + paca[cpu].default_decr = tb_ticks_per_jiffy; if (!cpu_has_feature(CPU_FTR_SLB)) { void *tmp; diff --git a/arch/ppc64/kernel/sysfs.c b/arch/ppc64/kernel/sysfs.c index 0925694c3ce5..c8fa6569b2fd 100644 --- a/arch/ppc64/kernel/sysfs.c +++ b/arch/ppc64/kernel/sysfs.c @@ -113,7 +113,6 @@ void ppc64_enable_pmcs(void) #ifdef CONFIG_PPC_PSERIES unsigned long set, reset; int ret; - unsigned int ctrl; #endif /* CONFIG_PPC_PSERIES */ /* Only need to enable them once */ @@ -167,11 +166,8 @@ void ppc64_enable_pmcs(void) * On SMT machines we have to set the run latch in the ctrl register * in order to make PMC6 spin. */ - if (cpu_has_feature(CPU_FTR_SMT)) { - ctrl = mfspr(CTRLF); - ctrl |= RUNLATCH; - mtspr(CTRLT, ctrl); - } + if (cpu_has_feature(CPU_FTR_SMT)) + ppc64_runlatch_on(); #endif /* CONFIG_PPC_PSERIES */ } diff --git a/arch/s390/appldata/appldata_base.c b/arch/s390/appldata/appldata_base.c index 01ae1964c938..c067435bae45 100644 --- a/arch/s390/appldata/appldata_base.c +++ b/arch/s390/appldata/appldata_base.c @@ -28,6 +28,7 @@ //#include <linux/kernel_stat.h> #include <linux/notifier.h> #include <linux/cpu.h> +#include <linux/workqueue.h> #include "appldata.h" @@ -133,9 +134,12 @@ static int appldata_interval = APPLDATA_CPU_INTERVAL; static int appldata_timer_active; /* - * Tasklet + * Work queue */ -static struct tasklet_struct appldata_tasklet_struct; +static struct workqueue_struct *appldata_wq; +static void appldata_work_fn(void *data); +static DECLARE_WORK(appldata_work, appldata_work_fn, NULL); + /* * Ops list @@ -144,11 +148,11 @@ static DEFINE_SPINLOCK(appldata_ops_lock); static LIST_HEAD(appldata_ops_list); -/************************* timer, tasklet, DIAG ******************************/ +/*************************** timer, work, DIAG *******************************/ /* * appldata_timer_function() * - * schedule tasklet and reschedule timer + * schedule work and reschedule timer */ static void appldata_timer_function(unsigned long data, struct pt_regs *regs) { @@ -157,22 +161,22 @@ static void appldata_timer_function(unsigned long data, struct pt_regs *regs) atomic_read(&appldata_expire_count)); if (atomic_dec_and_test(&appldata_expire_count)) { atomic_set(&appldata_expire_count, num_online_cpus()); - tasklet_schedule((struct tasklet_struct *) data); + queue_work(appldata_wq, (struct work_struct *) data); } } /* - * appldata_tasklet_function() + * appldata_work_fn() * * call data gathering function for each (active) module */ -static void appldata_tasklet_function(unsigned long data) +static void appldata_work_fn(void *data) { struct list_head *lh; struct appldata_ops *ops; int i; - P_DEBUG(" -= Tasklet =-\n"); + P_DEBUG(" -= Work Queue =-\n"); i = 0; spin_lock(&appldata_ops_lock); list_for_each(lh, &appldata_ops_list) { @@ -231,7 +235,7 @@ static int appldata_diag(char record_nr, u16 function, unsigned long buffer, : "=d" (ry) : "d" (&(appldata_parameter_list)) : "cc"); return (int) ry; } -/********************** timer, tasklet, DIAG <END> ***************************/ +/************************ timer, work, DIAG <END> ****************************/ /****************************** /proc stuff **********************************/ @@ -411,7 +415,7 @@ appldata_generic_handler(ctl_table *ctl, int write, struct file *filp, struct list_head *lh; found = 0; - spin_lock_bh(&appldata_ops_lock); + spin_lock(&appldata_ops_lock); list_for_each(lh, &appldata_ops_list) { tmp_ops = list_entry(lh, struct appldata_ops, list); if (&tmp_ops->ctl_table[2] == ctl) { @@ -419,15 +423,15 @@ appldata_generic_handler(ctl_table *ctl, int write, struct file *filp, } } if (!found) { - spin_unlock_bh(&appldata_ops_lock); + spin_unlock(&appldata_ops_lock); return -ENODEV; } ops = ctl->data; if (!try_module_get(ops->owner)) { // protect this function - spin_unlock_bh(&appldata_ops_lock); + spin_unlock(&appldata_ops_lock); return -ENODEV; } - spin_unlock_bh(&appldata_ops_lock); + spin_unlock(&appldata_ops_lock); if (!*lenp || *ppos) { *lenp = 0; @@ -451,10 +455,11 @@ appldata_generic_handler(ctl_table *ctl, int write, struct file *filp, return -EFAULT; } - spin_lock_bh(&appldata_ops_lock); + spin_lock(&appldata_ops_lock); if ((buf[0] == '1') && (ops->active == 0)) { - if (!try_module_get(ops->owner)) { // protect tasklet - spin_unlock_bh(&appldata_ops_lock); + // protect work queue callback + if (!try_module_get(ops->owner)) { + spin_unlock(&appldata_ops_lock); module_put(ops->owner); return -ENODEV; } @@ -485,7 +490,7 @@ appldata_generic_handler(ctl_table *ctl, int write, struct file *filp, } module_put(ops->owner); } - spin_unlock_bh(&appldata_ops_lock); + spin_unlock(&appldata_ops_lock); out: *lenp = len; *ppos += len; @@ -529,7 +534,7 @@ int appldata_register_ops(struct appldata_ops *ops) } memset(ops->ctl_table, 0, 4*sizeof(struct ctl_table)); - spin_lock_bh(&appldata_ops_lock); + spin_lock(&appldata_ops_lock); list_for_each(lh, &appldata_ops_list) { tmp_ops = list_entry(lh, struct appldata_ops, list); P_DEBUG("register_ops loop: %i) name = %s, ctl = %i\n", @@ -541,18 +546,18 @@ int appldata_register_ops(struct appldata_ops *ops) APPLDATA_PROC_NAME_LENGTH) == 0) { P_ERROR("Name \"%s\" already registered!\n", ops->name); kfree(ops->ctl_table); - spin_unlock_bh(&appldata_ops_lock); + spin_unlock(&appldata_ops_lock); return -EBUSY; } if (tmp_ops->ctl_nr == ops->ctl_nr) { P_ERROR("ctl_nr %i already registered!\n", ops->ctl_nr); kfree(ops->ctl_table); - spin_unlock_bh(&appldata_ops_lock); + spin_unlock(&appldata_ops_lock); return -EBUSY; } } list_add(&ops->list, &appldata_ops_list); - spin_unlock_bh(&appldata_ops_lock); + spin_unlock(&appldata_ops_lock); ops->ctl_table[0].ctl_name = CTL_APPLDATA; ops->ctl_table[0].procname = appldata_proc_name; @@ -583,12 +588,12 @@ int appldata_register_ops(struct appldata_ops *ops) */ void appldata_unregister_ops(struct appldata_ops *ops) { - spin_lock_bh(&appldata_ops_lock); + spin_lock(&appldata_ops_lock); unregister_sysctl_table(ops->sysctl_header); list_del(&ops->list); kfree(ops->ctl_table); ops->ctl_table = NULL; - spin_unlock_bh(&appldata_ops_lock); + spin_unlock(&appldata_ops_lock); P_INFO("%s-ops unregistered!\n", ops->name); } /********************** module-ops management <END> **************************/ @@ -602,7 +607,7 @@ appldata_online_cpu(int cpu) init_virt_timer(&per_cpu(appldata_timer, cpu)); per_cpu(appldata_timer, cpu).function = appldata_timer_function; per_cpu(appldata_timer, cpu).data = (unsigned long) - &appldata_tasklet_struct; + &appldata_work; atomic_inc(&appldata_expire_count); spin_lock(&appldata_timer_lock); __appldata_vtimer_setup(APPLDATA_MOD_TIMER); @@ -615,7 +620,7 @@ appldata_offline_cpu(int cpu) del_virt_timer(&per_cpu(appldata_timer, cpu)); if (atomic_dec_and_test(&appldata_expire_count)) { atomic_set(&appldata_expire_count, num_online_cpus()); - tasklet_schedule(&appldata_tasklet_struct); + queue_work(appldata_wq, &appldata_work); } spin_lock(&appldata_timer_lock); __appldata_vtimer_setup(APPLDATA_MOD_TIMER); @@ -648,7 +653,7 @@ static struct notifier_block __devinitdata appldata_nb = { /* * appldata_init() * - * init timer and tasklet, register /proc entries + * init timer, register /proc entries */ static int __init appldata_init(void) { @@ -657,6 +662,12 @@ static int __init appldata_init(void) P_DEBUG("sizeof(parameter_list) = %lu\n", sizeof(struct appldata_parameter_list)); + appldata_wq = create_singlethread_workqueue("appldata"); + if (!appldata_wq) { + P_ERROR("Could not create work queue\n"); + return -ENOMEM; + } + for_each_online_cpu(i) appldata_online_cpu(i); @@ -670,7 +681,6 @@ static int __init appldata_init(void) appldata_table[1].de->owner = THIS_MODULE; #endif - tasklet_init(&appldata_tasklet_struct, appldata_tasklet_function, 0); P_DEBUG("Base interface initialized.\n"); return 0; } @@ -678,7 +688,7 @@ static int __init appldata_init(void) /* * appldata_exit() * - * stop timer and tasklet, unregister /proc entries + * stop timer, unregister /proc entries */ static void __exit appldata_exit(void) { @@ -690,7 +700,7 @@ static void __exit appldata_exit(void) /* * ops list should be empty, but just in case something went wrong... */ - spin_lock_bh(&appldata_ops_lock); + spin_lock(&appldata_ops_lock); list_for_each(lh, &appldata_ops_list) { ops = list_entry(lh, struct appldata_ops, list); rc = appldata_diag(ops->record_nr, APPLDATA_STOP_REC, @@ -700,7 +710,7 @@ static void __exit appldata_exit(void) "return code: %d\n", ops->name, rc); } } - spin_unlock_bh(&appldata_ops_lock); + spin_unlock(&appldata_ops_lock); for_each_online_cpu(i) appldata_offline_cpu(i); @@ -709,7 +719,7 @@ static void __exit appldata_exit(void) unregister_sysctl_table(appldata_sysctl_header); - tasklet_kill(&appldata_tasklet_struct); + destroy_workqueue(appldata_wq); P_DEBUG("... module unloaded!\n"); } /**************************** init / exit <END> ******************************/ diff --git a/arch/s390/appldata/appldata_mem.c b/arch/s390/appldata/appldata_mem.c index 462ee9a84e76..f0e2fbed3d4c 100644 --- a/arch/s390/appldata/appldata_mem.c +++ b/arch/s390/appldata/appldata_mem.c @@ -68,7 +68,7 @@ struct appldata_mem_data { u64 pgmajfault; /* page faults (major only) */ // <-- New in 2.6 -} appldata_mem_data; +} __attribute__((packed)) appldata_mem_data; static inline void appldata_debug_print(struct appldata_mem_data *mem_data) diff --git a/arch/s390/appldata/appldata_net_sum.c b/arch/s390/appldata/appldata_net_sum.c index dd61638d3027..2a4c7432db4a 100644 --- a/arch/s390/appldata/appldata_net_sum.c +++ b/arch/s390/appldata/appldata_net_sum.c @@ -57,7 +57,7 @@ struct appldata_net_sum_data { u64 rx_dropped; /* no space in linux buffers */ u64 tx_dropped; /* no space available in linux */ u64 collisions; /* collisions while transmitting */ -} appldata_net_sum_data; +} __attribute__((packed)) appldata_net_sum_data; static inline void appldata_print_debug(struct appldata_net_sum_data *net_data) diff --git a/arch/s390/appldata/appldata_os.c b/arch/s390/appldata/appldata_os.c index b83f07484551..e0a476bf4fd6 100644 --- a/arch/s390/appldata/appldata_os.c +++ b/arch/s390/appldata/appldata_os.c @@ -49,7 +49,7 @@ struct appldata_os_per_cpu { u32 per_cpu_softirq; /* ... spent in softirqs */ u32 per_cpu_iowait; /* ... spent while waiting for I/O */ // <-- New in 2.6 -}; +} __attribute__((packed)); struct appldata_os_data { u64 timestamp; @@ -75,7 +75,7 @@ struct appldata_os_data { /* per cpu data */ struct appldata_os_per_cpu os_cpu[0]; -}; +} __attribute__((packed)); static struct appldata_os_data *appldata_os_data; diff --git a/arch/s390/kernel/ptrace.c b/arch/s390/kernel/ptrace.c index 26889366929a..06afa3103ace 100644 --- a/arch/s390/kernel/ptrace.c +++ b/arch/s390/kernel/ptrace.c @@ -40,6 +40,7 @@ #include <asm/pgalloc.h> #include <asm/system.h> #include <asm/uaccess.h> +#include <asm/unistd.h> #ifdef CONFIG_S390_SUPPORT #include "compat_ptrace.h" @@ -130,13 +131,19 @@ static int peek_user(struct task_struct *child, addr_t addr, addr_t data) { struct user *dummy = NULL; - addr_t offset, tmp; + addr_t offset, tmp, mask; /* * Stupid gdb peeks/pokes the access registers in 64 bit with * an alignment of 4. Programmers from hell... */ - if ((addr & 3) || addr > sizeof(struct user) - __ADDR_MASK) + mask = __ADDR_MASK; +#ifdef CONFIG_ARCH_S390X + if (addr >= (addr_t) &dummy->regs.acrs && + addr < (addr_t) &dummy->regs.orig_gpr2) + mask = 3; +#endif + if ((addr & mask) || addr > sizeof(struct user) - __ADDR_MASK) return -EIO; if (addr < (addr_t) &dummy->regs.acrs) { @@ -153,6 +160,16 @@ peek_user(struct task_struct *child, addr_t addr, addr_t data) * access registers are stored in the thread structure */ offset = addr - (addr_t) &dummy->regs.acrs; +#ifdef CONFIG_ARCH_S390X + /* + * Very special case: old & broken 64 bit gdb reading + * from acrs[15]. Result is a 64 bit value. Read the + * 32 bit acrs[15] value and shift it by 32. Sick... + */ + if (addr == (addr_t) &dummy->regs.acrs[15]) + tmp = ((unsigned long) child->thread.acrs[15]) << 32; + else +#endif tmp = *(addr_t *)((addr_t) &child->thread.acrs + offset); } else if (addr == (addr_t) &dummy->regs.orig_gpr2) { @@ -167,6 +184,9 @@ peek_user(struct task_struct *child, addr_t addr, addr_t data) */ offset = addr - (addr_t) &dummy->regs.fp_regs; tmp = *(addr_t *)((addr_t) &child->thread.fp_regs + offset); + if (addr == (addr_t) &dummy->regs.fp_regs.fpc) + tmp &= (unsigned long) FPC_VALID_MASK + << (BITS_PER_LONG - 32); } else if (addr < (addr_t) (&dummy->regs.per_info + 1)) { /* @@ -191,13 +211,19 @@ static int poke_user(struct task_struct *child, addr_t addr, addr_t data) { struct user *dummy = NULL; - addr_t offset; + addr_t offset, mask; /* * Stupid gdb peeks/pokes the access registers in 64 bit with * an alignment of 4. Programmers from hell indeed... */ - if ((addr & 3) || addr > sizeof(struct user) - __ADDR_MASK) + mask = __ADDR_MASK; +#ifdef CONFIG_ARCH_S390X + if (addr >= (addr_t) &dummy->regs.acrs && + addr < (addr_t) &dummy->regs.orig_gpr2) + mask = 3; +#endif + if ((addr & mask) || addr > sizeof(struct user) - __ADDR_MASK) return -EIO; if (addr < (addr_t) &dummy->regs.acrs) { @@ -224,6 +250,17 @@ poke_user(struct task_struct *child, addr_t addr, addr_t data) * access registers are stored in the thread structure */ offset = addr - (addr_t) &dummy->regs.acrs; +#ifdef CONFIG_ARCH_S390X + /* + * Very special case: old & broken 64 bit gdb writing + * to acrs[15] with a 64 bit value. Ignore the lower + * half of the value and write the upper 32 bit to + * acrs[15]. Sick... + */ + if (addr == (addr_t) &dummy->regs.acrs[15]) + child->thread.acrs[15] = (unsigned int) (data >> 32); + else +#endif *(addr_t *)((addr_t) &child->thread.acrs + offset) = data; } else if (addr == (addr_t) &dummy->regs.orig_gpr2) { @@ -237,7 +274,8 @@ poke_user(struct task_struct *child, addr_t addr, addr_t data) * floating point regs. are stored in the thread structure */ if (addr == (addr_t) &dummy->regs.fp_regs.fpc && - (data & ~FPC_VALID_MASK) != 0) + (data & ~((unsigned long) FPC_VALID_MASK + << (BITS_PER_LONG - 32))) != 0) return -EINVAL; offset = addr - (addr_t) &dummy->regs.fp_regs; *(addr_t *)((addr_t) &child->thread.fp_regs + offset) = data; @@ -723,6 +761,13 @@ syscall_trace(struct pt_regs *regs, int entryexit) ? 0x80 : 0)); /* + * If the debuffer has set an invalid system call number, + * we prepare to skip the system call restart handling. + */ + if (!entryexit && regs->gprs[2] >= NR_syscalls) + regs->trap = -1; + + /* * this isn't the same as continuing with a signal, but it will do * for normal use. strace only continues with a signal if the * stopping signal is not SIGTRAP. -brl diff --git a/arch/s390/mm/fault.c b/arch/s390/mm/fault.c index 80306bc8c799..75fde949d125 100644 --- a/arch/s390/mm/fault.c +++ b/arch/s390/mm/fault.c @@ -207,7 +207,7 @@ do_exception(struct pt_regs *regs, unsigned long error_code, int is_protection) * we are not in an interrupt and that there is a * user context. */ - if (user_address == 0 || in_interrupt() || !mm) + if (user_address == 0 || in_atomic() || !mm) goto no_context; /* diff --git a/arch/sparc64/kernel/pci_iommu.c b/arch/sparc64/kernel/pci_iommu.c index 33ca56c90da2..2803bc7c2c79 100644 --- a/arch/sparc64/kernel/pci_iommu.c +++ b/arch/sparc64/kernel/pci_iommu.c @@ -196,6 +196,34 @@ static iopte_t *alloc_consistent_cluster(struct pci_iommu *iommu, unsigned long return NULL; } +static int iommu_alloc_ctx(struct pci_iommu *iommu) +{ + int lowest = iommu->ctx_lowest_free; + int sz = IOMMU_NUM_CTXS - lowest; + int n = find_next_zero_bit(iommu->ctx_bitmap, sz, lowest); + + if (unlikely(n == sz)) { + n = find_next_zero_bit(iommu->ctx_bitmap, lowest, 1); + if (unlikely(n == lowest)) { + printk(KERN_WARNING "IOMMU: Ran out of contexts.\n"); + n = 0; + } + } + if (n) + __set_bit(n, iommu->ctx_bitmap); + + return n; +} + +static inline void iommu_free_ctx(struct pci_iommu *iommu, int ctx) +{ + if (likely(ctx)) { + __clear_bit(ctx, iommu->ctx_bitmap); + if (ctx < iommu->ctx_lowest_free) + iommu->ctx_lowest_free = ctx; + } +} + /* Allocate and map kernel buffer of size SIZE using consistent mode * DMA for PCI device PDEV. Return non-NULL cpu-side address if * successful and set *DMA_ADDRP to the PCI side dma address. @@ -236,7 +264,7 @@ void *pci_alloc_consistent(struct pci_dev *pdev, size_t size, dma_addr_t *dma_ad npages = size >> IO_PAGE_SHIFT; ctx = 0; if (iommu->iommu_ctxflush) - ctx = iommu->iommu_cur_ctx++; + ctx = iommu_alloc_ctx(iommu); first_page = __pa(first_page); while (npages--) { iopte_val(*iopte) = (IOPTE_CONSISTENT(ctx) | @@ -317,6 +345,8 @@ void pci_free_consistent(struct pci_dev *pdev, size_t size, void *cpu, dma_addr_ } } + iommu_free_ctx(iommu, ctx); + spin_unlock_irqrestore(&iommu->lock, flags); order = get_order(size); @@ -360,7 +390,7 @@ dma_addr_t pci_map_single(struct pci_dev *pdev, void *ptr, size_t sz, int direct base_paddr = __pa(oaddr & IO_PAGE_MASK); ctx = 0; if (iommu->iommu_ctxflush) - ctx = iommu->iommu_cur_ctx++; + ctx = iommu_alloc_ctx(iommu); if (strbuf->strbuf_enabled) iopte_protection = IOPTE_STREAMING(ctx); else @@ -380,39 +410,53 @@ bad: return PCI_DMA_ERROR_CODE; } -static void pci_strbuf_flush(struct pci_strbuf *strbuf, struct pci_iommu *iommu, u32 vaddr, unsigned long ctx, unsigned long npages) +static void pci_strbuf_flush(struct pci_strbuf *strbuf, struct pci_iommu *iommu, u32 vaddr, unsigned long ctx, unsigned long npages, int direction) { int limit; - PCI_STC_FLUSHFLAG_INIT(strbuf); if (strbuf->strbuf_ctxflush && iommu->iommu_ctxflush) { unsigned long matchreg, flushreg; + u64 val; flushreg = strbuf->strbuf_ctxflush; matchreg = PCI_STC_CTXMATCH_ADDR(strbuf, ctx); - limit = 100000; pci_iommu_write(flushreg, ctx); - for(;;) { - if (((long)pci_iommu_read(matchreg)) >= 0L) - break; - limit--; - if (!limit) - break; - udelay(1); + val = pci_iommu_read(matchreg); + val &= 0xffff; + if (!val) + goto do_flush_sync; + + while (val) { + if (val & 0x1) + pci_iommu_write(flushreg, ctx); + val >>= 1; } - if (!limit) + val = pci_iommu_read(matchreg); + if (unlikely(val)) { printk(KERN_WARNING "pci_strbuf_flush: ctx flush " - "timeout vaddr[%08x] ctx[%lx]\n", - vaddr, ctx); + "timeout matchreg[%lx] ctx[%lx]\n", + val, ctx); + goto do_page_flush; + } } else { unsigned long i; + do_page_flush: for (i = 0; i < npages; i++, vaddr += IO_PAGE_SIZE) pci_iommu_write(strbuf->strbuf_pflush, vaddr); } +do_flush_sync: + /* If the device could not have possibly put dirty data into + * the streaming cache, no flush-flag synchronization needs + * to be performed. + */ + if (direction == PCI_DMA_TODEVICE) + return; + + PCI_STC_FLUSHFLAG_INIT(strbuf); pci_iommu_write(strbuf->strbuf_fsync, strbuf->strbuf_flushflag_pa); (void) pci_iommu_read(iommu->write_complete_reg); @@ -466,7 +510,7 @@ void pci_unmap_single(struct pci_dev *pdev, dma_addr_t bus_addr, size_t sz, int /* Step 1: Kick data out of streaming buffers if necessary. */ if (strbuf->strbuf_enabled) - pci_strbuf_flush(strbuf, iommu, bus_addr, ctx, npages); + pci_strbuf_flush(strbuf, iommu, bus_addr, ctx, npages, direction); /* Step 2: Clear out first TSB entry. */ iopte_make_dummy(iommu, base); @@ -474,6 +518,8 @@ void pci_unmap_single(struct pci_dev *pdev, dma_addr_t bus_addr, size_t sz, int free_streaming_cluster(iommu, bus_addr - iommu->page_table_map_base, npages, ctx); + iommu_free_ctx(iommu, ctx); + spin_unlock_irqrestore(&iommu->lock, flags); } @@ -613,7 +659,7 @@ int pci_map_sg(struct pci_dev *pdev, struct scatterlist *sglist, int nelems, int /* Step 4: Choose a context if necessary. */ ctx = 0; if (iommu->iommu_ctxflush) - ctx = iommu->iommu_cur_ctx++; + ctx = iommu_alloc_ctx(iommu); /* Step 5: Create the mappings. */ if (strbuf->strbuf_enabled) @@ -678,7 +724,7 @@ void pci_unmap_sg(struct pci_dev *pdev, struct scatterlist *sglist, int nelems, /* Step 1: Kick data out of streaming buffers if necessary. */ if (strbuf->strbuf_enabled) - pci_strbuf_flush(strbuf, iommu, bus_addr, ctx, npages); + pci_strbuf_flush(strbuf, iommu, bus_addr, ctx, npages, direction); /* Step 2: Clear out first TSB entry. */ iopte_make_dummy(iommu, base); @@ -686,6 +732,8 @@ void pci_unmap_sg(struct pci_dev *pdev, struct scatterlist *sglist, int nelems, free_streaming_cluster(iommu, bus_addr - iommu->page_table_map_base, npages, ctx); + iommu_free_ctx(iommu, ctx); + spin_unlock_irqrestore(&iommu->lock, flags); } @@ -724,7 +772,7 @@ void pci_dma_sync_single_for_cpu(struct pci_dev *pdev, dma_addr_t bus_addr, size } /* Step 2: Kick data out of streaming buffers. */ - pci_strbuf_flush(strbuf, iommu, bus_addr, ctx, npages); + pci_strbuf_flush(strbuf, iommu, bus_addr, ctx, npages, direction); spin_unlock_irqrestore(&iommu->lock, flags); } @@ -768,7 +816,7 @@ void pci_dma_sync_sg_for_cpu(struct pci_dev *pdev, struct scatterlist *sglist, i i--; npages = (IO_PAGE_ALIGN(sglist[i].dma_address + sglist[i].dma_length) - bus_addr) >> IO_PAGE_SHIFT; - pci_strbuf_flush(strbuf, iommu, bus_addr, ctx, npages); + pci_strbuf_flush(strbuf, iommu, bus_addr, ctx, npages, direction); spin_unlock_irqrestore(&iommu->lock, flags); } diff --git a/arch/sparc64/kernel/pci_psycho.c b/arch/sparc64/kernel/pci_psycho.c index 3567fa879e1f..534320ef0db2 100644 --- a/arch/sparc64/kernel/pci_psycho.c +++ b/arch/sparc64/kernel/pci_psycho.c @@ -1212,7 +1212,7 @@ static void __init psycho_iommu_init(struct pci_controller_info *p) /* Setup initial software IOMMU state. */ spin_lock_init(&iommu->lock); - iommu->iommu_cur_ctx = 0; + iommu->ctx_lowest_free = 1; /* Register addresses. */ iommu->iommu_control = p->pbm_A.controller_regs + PSYCHO_IOMMU_CONTROL; diff --git a/arch/sparc64/kernel/pci_sabre.c b/arch/sparc64/kernel/pci_sabre.c index 5525d1ec4af8..53d333b4a4e8 100644 --- a/arch/sparc64/kernel/pci_sabre.c +++ b/arch/sparc64/kernel/pci_sabre.c @@ -1265,7 +1265,7 @@ static void __init sabre_iommu_init(struct pci_controller_info *p, /* Setup initial software IOMMU state. */ spin_lock_init(&iommu->lock); - iommu->iommu_cur_ctx = 0; + iommu->ctx_lowest_free = 1; /* Register addresses. */ iommu->iommu_control = p->pbm_A.controller_regs + SABRE_IOMMU_CONTROL; diff --git a/arch/sparc64/kernel/pci_schizo.c b/arch/sparc64/kernel/pci_schizo.c index e93fcadc3722..5753175b94e6 100644 --- a/arch/sparc64/kernel/pci_schizo.c +++ b/arch/sparc64/kernel/pci_schizo.c @@ -1753,7 +1753,7 @@ static void schizo_pbm_iommu_init(struct pci_pbm_info *pbm) /* Setup initial software IOMMU state. */ spin_lock_init(&iommu->lock); - iommu->iommu_cur_ctx = 0; + iommu->ctx_lowest_free = 1; /* Register addresses, SCHIZO has iommu ctx flushing. */ iommu->iommu_control = pbm->pbm_regs + SCHIZO_IOMMU_CONTROL; diff --git a/arch/sparc64/kernel/sbus.c b/arch/sparc64/kernel/sbus.c index 76ea6455433f..89f5e019f24c 100644 --- a/arch/sparc64/kernel/sbus.c +++ b/arch/sparc64/kernel/sbus.c @@ -117,17 +117,25 @@ static void iommu_flush(struct sbus_iommu *iommu, u32 base, unsigned long npages #define STRBUF_TAG_VALID 0x02UL -static void sbus_strbuf_flush(struct sbus_iommu *iommu, u32 base, unsigned long npages) +static void sbus_strbuf_flush(struct sbus_iommu *iommu, u32 base, unsigned long npages, int direction) { unsigned long n; int limit; - iommu->strbuf_flushflag = 0UL; n = npages; while (n--) upa_writeq(base + (n << IO_PAGE_SHIFT), iommu->strbuf_regs + STRBUF_PFLUSH); + /* If the device could not have possibly put dirty data into + * the streaming cache, no flush-flag synchronization needs + * to be performed. + */ + if (direction == SBUS_DMA_TODEVICE) + return; + + iommu->strbuf_flushflag = 0UL; + /* Whoopee cushion! */ upa_writeq(__pa(&iommu->strbuf_flushflag), iommu->strbuf_regs + STRBUF_FSYNC); @@ -421,7 +429,7 @@ void sbus_unmap_single(struct sbus_dev *sdev, dma_addr_t dma_addr, size_t size, spin_lock_irqsave(&iommu->lock, flags); free_streaming_cluster(iommu, dma_base, size >> IO_PAGE_SHIFT); - sbus_strbuf_flush(iommu, dma_base, size >> IO_PAGE_SHIFT); + sbus_strbuf_flush(iommu, dma_base, size >> IO_PAGE_SHIFT, direction); spin_unlock_irqrestore(&iommu->lock, flags); } @@ -584,7 +592,7 @@ void sbus_unmap_sg(struct sbus_dev *sdev, struct scatterlist *sg, int nents, int iommu = sdev->bus->iommu; spin_lock_irqsave(&iommu->lock, flags); free_streaming_cluster(iommu, dvma_base, size >> IO_PAGE_SHIFT); - sbus_strbuf_flush(iommu, dvma_base, size >> IO_PAGE_SHIFT); + sbus_strbuf_flush(iommu, dvma_base, size >> IO_PAGE_SHIFT, direction); spin_unlock_irqrestore(&iommu->lock, flags); } @@ -596,7 +604,7 @@ void sbus_dma_sync_single_for_cpu(struct sbus_dev *sdev, dma_addr_t base, size_t size = (IO_PAGE_ALIGN(base + size) - (base & IO_PAGE_MASK)); spin_lock_irqsave(&iommu->lock, flags); - sbus_strbuf_flush(iommu, base & IO_PAGE_MASK, size >> IO_PAGE_SHIFT); + sbus_strbuf_flush(iommu, base & IO_PAGE_MASK, size >> IO_PAGE_SHIFT, direction); spin_unlock_irqrestore(&iommu->lock, flags); } @@ -620,7 +628,7 @@ void sbus_dma_sync_sg_for_cpu(struct sbus_dev *sdev, struct scatterlist *sg, int size = IO_PAGE_ALIGN(sg[i].dma_address + sg[i].dma_length) - base; spin_lock_irqsave(&iommu->lock, flags); - sbus_strbuf_flush(iommu, base, size >> IO_PAGE_SHIFT); + sbus_strbuf_flush(iommu, base, size >> IO_PAGE_SHIFT, direction); spin_unlock_irqrestore(&iommu->lock, flags); } |