From 08280e6c4c2e8049ac61d9e8e3536ec1df629c0d Mon Sep 17 00:00:00 2001 From: "David S. Miller" Date: Sun, 14 Oct 2012 17:59:40 -0700 Subject: sparc64: Like x86 we should check current->mm during perf backtrace generation. If the MM is not active, only report the top-level PC. Do not try to access the address space. Signed-off-by: David S. Miller --- arch/sparc/kernel/perf_event.c | 9 +++++---- 1 file changed, 5 insertions(+), 4 deletions(-) (limited to 'arch/sparc') diff --git a/arch/sparc/kernel/perf_event.c b/arch/sparc/kernel/perf_event.c index e48651dace1b..9e96f849a744 100644 --- a/arch/sparc/kernel/perf_event.c +++ b/arch/sparc/kernel/perf_event.c @@ -1738,8 +1738,6 @@ static void perf_callchain_user_64(struct perf_callchain_entry *entry, { unsigned long ufp; - perf_callchain_store(entry, regs->tpc); - ufp = regs->u_regs[UREG_I6] + STACK_BIAS; do { struct sparc_stackf *usf, sf; @@ -1760,8 +1758,6 @@ static void perf_callchain_user_32(struct perf_callchain_entry *entry, { unsigned long ufp; - perf_callchain_store(entry, regs->tpc); - ufp = regs->u_regs[UREG_I6] & 0xffffffffUL; do { struct sparc_stackf32 *usf, sf; @@ -1780,6 +1776,11 @@ static void perf_callchain_user_32(struct perf_callchain_entry *entry, void perf_callchain_user(struct perf_callchain_entry *entry, struct pt_regs *regs) { + perf_callchain_store(entry, regs->tpc); + + if (!current->mm) + return; + flushw_user(); if (test_thread_flag(TIF_32BIT)) perf_callchain_user_32(entry, regs); -- cgit v1.2.3 From 916ca14aaf12a7191118adb51bb95e3c7866380d Mon Sep 17 00:00:00 2001 From: "David S. Miller" Date: Tue, 16 Oct 2012 09:34:01 -0700 Subject: sparc64: Add global PMU register dumping via sysrq. Signed-off-by: David S. Miller --- arch/sparc/include/asm/ptrace.h | 13 ++++- arch/sparc/include/asm/smp_64.h | 2 + arch/sparc/kernel/process_64.c | 120 +++++++++++++++++++++++++++++++++------- arch/sparc/kernel/smp_64.c | 11 ++++ arch/sparc/mm/ultra.S | 64 ++++++++++++++++++++- 5 files changed, 187 insertions(+), 23 deletions(-) (limited to 'arch/sparc') diff --git a/arch/sparc/include/asm/ptrace.h b/arch/sparc/include/asm/ptrace.h index 0c6f6b068289..da43bdc62294 100644 --- a/arch/sparc/include/asm/ptrace.h +++ b/arch/sparc/include/asm/ptrace.h @@ -42,7 +42,18 @@ struct global_reg_snapshot { struct thread_info *thread; unsigned long pad1; }; -extern struct global_reg_snapshot global_reg_snapshot[NR_CPUS]; + +struct global_pmu_snapshot { + unsigned long pcr[4]; + unsigned long pic[4]; +}; + +union global_cpu_snapshot { + struct global_reg_snapshot reg; + struct global_pmu_snapshot pmu; +}; + +extern union global_cpu_snapshot global_cpu_snapshot[NR_CPUS]; #define force_successful_syscall_return() \ do { current_thread_info()->syscall_noerror = 1; \ diff --git a/arch/sparc/include/asm/smp_64.h b/arch/sparc/include/asm/smp_64.h index 29862a9e9065..dd3bef4b9896 100644 --- a/arch/sparc/include/asm/smp_64.h +++ b/arch/sparc/include/asm/smp_64.h @@ -48,6 +48,7 @@ extern void smp_fill_in_sib_core_maps(void); extern void cpu_play_dead(void); extern void smp_fetch_global_regs(void); +extern void smp_fetch_global_pmu(void); struct seq_file; void smp_bogo(struct seq_file *); @@ -65,6 +66,7 @@ extern void __cpu_die(unsigned int cpu); #define hard_smp_processor_id() 0 #define smp_fill_in_sib_core_maps() do { } while (0) #define smp_fetch_global_regs() do { } while (0) +#define smp_fetch_global_pmu() do { } while (0) #endif /* !(CONFIG_SMP) */ diff --git a/arch/sparc/kernel/process_64.c b/arch/sparc/kernel/process_64.c index fcaa59421126..d778248ef3f8 100644 --- a/arch/sparc/kernel/process_64.c +++ b/arch/sparc/kernel/process_64.c @@ -27,6 +27,7 @@ #include #include #include +#include #include #include #include @@ -47,6 +48,7 @@ #include #include #include +#include #include "kstack.h" @@ -204,18 +206,22 @@ void show_regs(struct pt_regs *regs) show_stack(current, (unsigned long *) regs->u_regs[UREG_FP]); } -struct global_reg_snapshot global_reg_snapshot[NR_CPUS]; -static DEFINE_SPINLOCK(global_reg_snapshot_lock); +union global_cpu_snapshot global_cpu_snapshot[NR_CPUS]; +static DEFINE_SPINLOCK(global_cpu_snapshot_lock); static void __global_reg_self(struct thread_info *tp, struct pt_regs *regs, int this_cpu) { + struct global_reg_snapshot *rp; + flushw_all(); - global_reg_snapshot[this_cpu].tstate = regs->tstate; - global_reg_snapshot[this_cpu].tpc = regs->tpc; - global_reg_snapshot[this_cpu].tnpc = regs->tnpc; - global_reg_snapshot[this_cpu].o7 = regs->u_regs[UREG_I7]; + rp = &global_cpu_snapshot[this_cpu].reg; + + rp->tstate = regs->tstate; + rp->tpc = regs->tpc; + rp->tnpc = regs->tnpc; + rp->o7 = regs->u_regs[UREG_I7]; if (regs->tstate & TSTATE_PRIV) { struct reg_window *rw; @@ -223,17 +229,17 @@ static void __global_reg_self(struct thread_info *tp, struct pt_regs *regs, rw = (struct reg_window *) (regs->u_regs[UREG_FP] + STACK_BIAS); if (kstack_valid(tp, (unsigned long) rw)) { - global_reg_snapshot[this_cpu].i7 = rw->ins[7]; + rp->i7 = rw->ins[7]; rw = (struct reg_window *) (rw->ins[6] + STACK_BIAS); if (kstack_valid(tp, (unsigned long) rw)) - global_reg_snapshot[this_cpu].rpc = rw->ins[7]; + rp->rpc = rw->ins[7]; } } else { - global_reg_snapshot[this_cpu].i7 = 0; - global_reg_snapshot[this_cpu].rpc = 0; + rp->i7 = 0; + rp->rpc = 0; } - global_reg_snapshot[this_cpu].thread = tp; + rp->thread = tp; } /* In order to avoid hangs we do not try to synchronize with the @@ -261,9 +267,9 @@ void arch_trigger_all_cpu_backtrace(void) if (!regs) regs = tp->kregs; - spin_lock_irqsave(&global_reg_snapshot_lock, flags); + spin_lock_irqsave(&global_cpu_snapshot_lock, flags); - memset(global_reg_snapshot, 0, sizeof(global_reg_snapshot)); + memset(global_cpu_snapshot, 0, sizeof(global_cpu_snapshot)); this_cpu = raw_smp_processor_id(); @@ -272,7 +278,7 @@ void arch_trigger_all_cpu_backtrace(void) smp_fetch_global_regs(); for_each_online_cpu(cpu) { - struct global_reg_snapshot *gp = &global_reg_snapshot[cpu]; + struct global_reg_snapshot *gp = &global_cpu_snapshot[cpu].reg; __global_reg_poll(gp); @@ -295,9 +301,9 @@ void arch_trigger_all_cpu_backtrace(void) } } - memset(global_reg_snapshot, 0, sizeof(global_reg_snapshot)); + memset(global_cpu_snapshot, 0, sizeof(global_cpu_snapshot)); - spin_unlock_irqrestore(&global_reg_snapshot_lock, flags); + spin_unlock_irqrestore(&global_cpu_snapshot_lock, flags); } #ifdef CONFIG_MAGIC_SYSRQ @@ -309,16 +315,90 @@ static void sysrq_handle_globreg(int key) static struct sysrq_key_op sparc_globalreg_op = { .handler = sysrq_handle_globreg, - .help_msg = "Globalregs", + .help_msg = "global-regs(Y)", .action_msg = "Show Global CPU Regs", }; -static int __init sparc_globreg_init(void) +static void __global_pmu_self(int this_cpu) +{ + struct global_pmu_snapshot *pp; + int i, num; + + pp = &global_cpu_snapshot[this_cpu].pmu; + + num = 1; + if (tlb_type == hypervisor && + sun4v_chip_type >= SUN4V_CHIP_NIAGARA4) + num = 4; + + for (i = 0; i < num; i++) { + pp->pcr[i] = pcr_ops->read_pcr(i); + pp->pic[i] = pcr_ops->read_pic(i); + } +} + +static void __global_pmu_poll(struct global_pmu_snapshot *pp) +{ + int limit = 0; + + while (!pp->pcr[0] && ++limit < 100) { + barrier(); + udelay(1); + } +} + +static void pmu_snapshot_all_cpus(void) { - return register_sysrq_key('y', &sparc_globalreg_op); + unsigned long flags; + int this_cpu, cpu; + + spin_lock_irqsave(&global_cpu_snapshot_lock, flags); + + memset(global_cpu_snapshot, 0, sizeof(global_cpu_snapshot)); + + this_cpu = raw_smp_processor_id(); + + __global_pmu_self(this_cpu); + + smp_fetch_global_pmu(); + + for_each_online_cpu(cpu) { + struct global_pmu_snapshot *pp = &global_cpu_snapshot[cpu].pmu; + + __global_pmu_poll(pp); + + printk("%c CPU[%3d]: PCR[%08lx:%08lx:%08lx:%08lx] PIC[%08lx:%08lx:%08lx:%08lx]\n", + (cpu == this_cpu ? '*' : ' '), cpu, + pp->pcr[0], pp->pcr[1], pp->pcr[2], pp->pcr[3], + pp->pic[0], pp->pic[1], pp->pic[2], pp->pic[3]); + } + + memset(global_cpu_snapshot, 0, sizeof(global_cpu_snapshot)); + + spin_unlock_irqrestore(&global_cpu_snapshot_lock, flags); +} + +static void sysrq_handle_globpmu(int key) +{ + pmu_snapshot_all_cpus(); +} + +static struct sysrq_key_op sparc_globalpmu_op = { + .handler = sysrq_handle_globpmu, + .help_msg = "global-pmu(X)", + .action_msg = "Show Global PMU Regs", +}; + +static int __init sparc_sysrq_init(void) +{ + int ret = register_sysrq_key('y', &sparc_globalreg_op); + + if (!ret) + ret = register_sysrq_key('x', &sparc_globalpmu_op); + return ret; } -core_initcall(sparc_globreg_init); +core_initcall(sparc_sysrq_init); #endif diff --git a/arch/sparc/kernel/smp_64.c b/arch/sparc/kernel/smp_64.c index 781bcb10b8bd..d94b878577b7 100644 --- a/arch/sparc/kernel/smp_64.c +++ b/arch/sparc/kernel/smp_64.c @@ -852,6 +852,8 @@ extern unsigned long xcall_flush_tlb_mm; extern unsigned long xcall_flush_tlb_pending; extern unsigned long xcall_flush_tlb_kernel_range; extern unsigned long xcall_fetch_glob_regs; +extern unsigned long xcall_fetch_glob_pmu; +extern unsigned long xcall_fetch_glob_pmu_n4; extern unsigned long xcall_receive_signal; extern unsigned long xcall_new_mmu_context_version; #ifdef CONFIG_KGDB @@ -1000,6 +1002,15 @@ void smp_fetch_global_regs(void) smp_cross_call(&xcall_fetch_glob_regs, 0, 0, 0); } +void smp_fetch_global_pmu(void) +{ + if (tlb_type == hypervisor && + sun4v_chip_type >= SUN4V_CHIP_NIAGARA4) + smp_cross_call(&xcall_fetch_glob_pmu_n4, 0, 0, 0); + else + smp_cross_call(&xcall_fetch_glob_pmu, 0, 0, 0); +} + /* We know that the window frames of the user have been flushed * to the stack before we get here because all callers of us * are flush_tlb_*() routines, and these run after flush_cache_*() diff --git a/arch/sparc/mm/ultra.S b/arch/sparc/mm/ultra.S index 874162a11ceb..f8e13d421fcb 100644 --- a/arch/sparc/mm/ultra.S +++ b/arch/sparc/mm/ultra.S @@ -481,8 +481,8 @@ xcall_sync_tick: .globl xcall_fetch_glob_regs xcall_fetch_glob_regs: - sethi %hi(global_reg_snapshot), %g1 - or %g1, %lo(global_reg_snapshot), %g1 + sethi %hi(global_cpu_snapshot), %g1 + or %g1, %lo(global_cpu_snapshot), %g1 __GET_CPUID(%g2) sllx %g2, 6, %g3 add %g1, %g3, %g1 @@ -509,6 +509,66 @@ xcall_fetch_glob_regs: stx %g3, [%g1 + GR_SNAP_THREAD] retry + .globl xcall_fetch_glob_pmu +xcall_fetch_glob_pmu: + sethi %hi(global_cpu_snapshot), %g1 + or %g1, %lo(global_cpu_snapshot), %g1 + __GET_CPUID(%g2) + sllx %g2, 6, %g3 + add %g1, %g3, %g1 + rd %pic, %g7 + stx %g7, [%g1 + (4 * 8)] + rd %pcr, %g7 + stx %g7, [%g1 + (0 * 8)] + retry + + .globl xcall_fetch_glob_pmu_n4 +xcall_fetch_glob_pmu_n4: + sethi %hi(global_cpu_snapshot), %g1 + or %g1, %lo(global_cpu_snapshot), %g1 + __GET_CPUID(%g2) + sllx %g2, 6, %g3 + add %g1, %g3, %g1 + + ldxa [%g0] ASI_PIC, %g7 + stx %g7, [%g1 + (4 * 8)] + mov 0x08, %g3 + ldxa [%g3] ASI_PIC, %g7 + stx %g7, [%g1 + (5 * 8)] + mov 0x10, %g3 + ldxa [%g3] ASI_PIC, %g7 + stx %g7, [%g1 + (6 * 8)] + mov 0x18, %g3 + ldxa [%g3] ASI_PIC, %g7 + stx %g7, [%g1 + (7 * 8)] + + mov %o0, %g2 + mov %o1, %g3 + mov %o5, %g7 + + mov HV_FAST_VT_GET_PERFREG, %o5 + mov 3, %o0 + ta HV_FAST_TRAP + stx %o1, [%g1 + (3 * 8)] + mov HV_FAST_VT_GET_PERFREG, %o5 + mov 2, %o0 + ta HV_FAST_TRAP + stx %o1, [%g1 + (2 * 8)] + mov HV_FAST_VT_GET_PERFREG, %o5 + mov 1, %o0 + ta HV_FAST_TRAP + stx %o1, [%g1 + (1 * 8)] + mov HV_FAST_VT_GET_PERFREG, %o5 + mov 0, %o0 + ta HV_FAST_TRAP + stx %o1, [%g1 + (0 * 8)] + + mov %g2, %o0 + mov %g3, %o1 + mov %g7, %o5 + + retry + #ifdef DCACHE_ALIASING_POSSIBLE .align 32 .globl xcall_flush_dcache_page_cheetah -- cgit v1.2.3 From e793d8c6740f8fe704fa216e95685f4d92c4c4b9 Mon Sep 17 00:00:00 2001 From: "David S. Miller" Date: Tue, 16 Oct 2012 13:05:25 -0700 Subject: sparc64: Fix bit twiddling in sparc_pmu_enable_event(). There was a serious disconnect in the logic happening in sparc_pmu_disable_event() vs. sparc_pmu_enable_event(). Event disable is implemented by programming a NOP event into the PCR. However, event enable was not reversing this operation. Instead, it was setting the User/Priv/Hypervisor trace enable bits. That's not sparc_pmu_enable_event()'s job, that's what sparc_pmu_enable() and sparc_pmu_disable() do . The intent of sparc_pmu_enable_event() is clear, since it first clear out the event type encoding field. So fix this by OR'ing in the event encoding rather than the trace enable bits. Signed-off-by: David S. Miller --- arch/sparc/kernel/perf_event.c | 6 ++++-- 1 file changed, 4 insertions(+), 2 deletions(-) (limited to 'arch/sparc') diff --git a/arch/sparc/kernel/perf_event.c b/arch/sparc/kernel/perf_event.c index 9e96f849a744..885a8af74064 100644 --- a/arch/sparc/kernel/perf_event.c +++ b/arch/sparc/kernel/perf_event.c @@ -817,15 +817,17 @@ static u64 nop_for_index(int idx) static inline void sparc_pmu_enable_event(struct cpu_hw_events *cpuc, struct hw_perf_event *hwc, int idx) { - u64 val, mask = mask_for_index(idx); + u64 enc, val, mask = mask_for_index(idx); int pcr_index = 0; if (sparc_pmu->num_pcrs > 1) pcr_index = idx; + enc = perf_event_get_enc(cpuc->events[idx]); + val = cpuc->pcr[pcr_index]; val &= ~mask; - val |= hwc->config; + val |= event_encoding(enc, idx); cpuc->pcr[pcr_index] = val; pcr_ops->write_pcr(pcr_index, cpuc->pcr[pcr_index]); -- cgit v1.2.3 From bb2bab177408e44079ba6bd37242fa8b26dfc2a7 Mon Sep 17 00:00:00 2001 From: David Howells Date: Wed, 17 Oct 2012 12:32:35 +0100 Subject: UAPI: Make arch/sparc/include/uapi/asm/sigcontext.h non-empty arch/sparc/include/uapi/asm/sigcontext.h was emitted by the UAPI disintegration script as an empty file because the parent file had no UAPI stuff in it, despite being marked with "header-y". Unfortunately, the patch program deletes resultant empty files when applying a kernel patch. So just stick a comment in there as a placeholder. Signed-off-by: David Howells cc: David S. Miller cc: sparclinux@vger.kernel.org --- arch/sparc/include/uapi/asm/sigcontext.h | 4 ++++ 1 file changed, 4 insertions(+) (limited to 'arch/sparc') diff --git a/arch/sparc/include/uapi/asm/sigcontext.h b/arch/sparc/include/uapi/asm/sigcontext.h index e69de29bb2d1..ae5704fa77ad 100644 --- a/arch/sparc/include/uapi/asm/sigcontext.h +++ b/arch/sparc/include/uapi/asm/sigcontext.h @@ -0,0 +1,4 @@ +/* + * There isn't anything here anymore, but the file must not be empty or patch + * will delete it. + */ -- cgit v1.2.3 From 517ffce4e1a03aea979fe3a18a3dd1761a24fafb Mon Sep 17 00:00:00 2001 From: "David S. Miller" Date: Fri, 26 Oct 2012 15:18:37 -0700 Subject: sparc64: Make montmul/montsqr/mpmul usable in 32-bit threads. The Montgomery Multiply, Montgomery Square, and Multiple-Precision Multiply instructions work by loading a combination of the floating point and multiple register windows worth of integer registers with the inputs. These values are 64-bit. But for 32-bit userland processes we only save the low 32-bits of each integer register during a register spill. This is because the register window save area is in the user stack and has a fixed layout. Therefore, the only way to use these instruction in 32-bit mode is to perform the following sequence: 1) Load the top-32bits of a choosen integer register with a sentinel, say "-1". This will be in the outer-most register window. The idea is that we're trying to see if the outer-most register window gets spilled, and thus the 64-bit values were truncated. 2) Load all the inputs for the montmul/montsqr/mpmul instruction, down to the inner-most register window. 3) Execute the opcode. 4) Traverse back up to the outer-most register window. 5) Check the sentinel, if it's still "-1" store the results. Otherwise retry the entire sequence. This retry is extremely troublesome. If you're just unlucky and an interrupt or other trap happens, it'll push that outer-most window to the stack and clear the sentinel when we restore it. We could retry forever and never make forward progress if interrupts arrive at a fast enough rate (consider perf events as one example). So we have do limited retries and fallback to software which is extremely non-deterministic. Luckily it's very straightforward to provide a mechanism to let 32-bit applications use a 64-bit stack. Stacks in 64-bit mode are biased by 2047 bytes, which means that the lowest bit is set in the actual %sp register value. So if we see bit zero set in a 32-bit application's stack we treat it like a 64-bit stack. Runtime detection of such a facility is tricky, and cumbersome at best. For example, just trying to use a biased stack and seeing if it works is hard to recover from (the signal handler will need to use an alt stack, plus something along the lines of longjmp). Therefore, we add a system call to report a bitmask of arch specific features like this in a cheap and less hairy way. With help from Andy Polyakov. Signed-off-by: David S. Miller --- arch/sparc/include/asm/compat.h | 5 ++-- arch/sparc/include/asm/thread_info_64.h | 5 ++++ arch/sparc/include/asm/ttable.h | 24 ++++++++++++------- arch/sparc/include/uapi/asm/unistd.h | 6 ++++- arch/sparc/kernel/perf_event.c | 22 ++++++++++++----- arch/sparc/kernel/process_64.c | 42 ++++++++++++++++++--------------- arch/sparc/kernel/ptrace_64.c | 4 ++-- arch/sparc/kernel/sys_sparc_64.c | 5 ++++ arch/sparc/kernel/systbls_64.S | 2 ++ arch/sparc/kernel/unaligned_64.c | 36 ++++++++++++++++++---------- arch/sparc/kernel/visemul.c | 23 +++++++++++------- arch/sparc/kernel/winfixup.S | 2 ++ arch/sparc/math-emu/math_64.c | 2 +- 13 files changed, 117 insertions(+), 61 deletions(-) (limited to 'arch/sparc') diff --git a/arch/sparc/include/asm/compat.h b/arch/sparc/include/asm/compat.h index cef99fbc0a21..830502fe62b4 100644 --- a/arch/sparc/include/asm/compat.h +++ b/arch/sparc/include/asm/compat.h @@ -232,9 +232,10 @@ static inline void __user *arch_compat_alloc_user_space(long len) struct pt_regs *regs = current_thread_info()->kregs; unsigned long usp = regs->u_regs[UREG_I6]; - if (!(test_thread_flag(TIF_32BIT))) + if (test_thread_64bit_stack(usp)) usp += STACK_BIAS; - else + + if (test_thread_flag(TIF_32BIT)) usp &= 0xffffffffUL; usp -= len; diff --git a/arch/sparc/include/asm/thread_info_64.h b/arch/sparc/include/asm/thread_info_64.h index 4e2276631081..a3fe4dcc0aa6 100644 --- a/arch/sparc/include/asm/thread_info_64.h +++ b/arch/sparc/include/asm/thread_info_64.h @@ -259,6 +259,11 @@ static inline bool test_and_clear_restore_sigmask(void) #define tsk_is_polling(t) test_tsk_thread_flag(t, TIF_POLLING_NRFLAG) +#define thread32_stack_is_64bit(__SP) (((__SP) & 0x1) != 0) +#define test_thread_64bit_stack(__SP) \ + ((test_thread_flag(TIF_32BIT) && !thread32_stack_is_64bit(__SP)) ? \ + false : true) + #endif /* !__ASSEMBLY__ */ #endif /* __KERNEL__ */ diff --git a/arch/sparc/include/asm/ttable.h b/arch/sparc/include/asm/ttable.h index 48f2807d3265..71b5a67522ab 100644 --- a/arch/sparc/include/asm/ttable.h +++ b/arch/sparc/include/asm/ttable.h @@ -372,7 +372,9 @@ etrap_spill_fixup_64bit: \ /* Normal 32bit spill */ #define SPILL_2_GENERIC(ASI) \ - srl %sp, 0, %sp; \ + and %sp, 1, %g3; \ + brnz,pn %g3, (. - (128 + 4)); \ + srl %sp, 0, %sp; \ stwa %l0, [%sp + %g0] ASI; \ mov 0x04, %g3; \ stwa %l1, [%sp + %g3] ASI; \ @@ -398,14 +400,16 @@ etrap_spill_fixup_64bit: \ stwa %i6, [%g1 + %g0] ASI; \ stwa %i7, [%g1 + %g3] ASI; \ saved; \ - retry; nop; nop; \ + retry; \ b,a,pt %xcc, spill_fixup_dax; \ b,a,pt %xcc, spill_fixup_mna; \ b,a,pt %xcc, spill_fixup; #define SPILL_2_GENERIC_ETRAP \ etrap_user_spill_32bit: \ - srl %sp, 0, %sp; \ + and %sp, 1, %g3; \ + brnz,pn %g3, etrap_user_spill_64bit; \ + srl %sp, 0, %sp; \ stwa %l0, [%sp + 0x00] %asi; \ stwa %l1, [%sp + 0x04] %asi; \ stwa %l2, [%sp + 0x08] %asi; \ @@ -427,7 +431,7 @@ etrap_user_spill_32bit: \ ba,pt %xcc, etrap_save; \ wrpr %g1, %cwp; \ nop; nop; nop; nop; \ - nop; nop; nop; nop; \ + nop; nop; \ ba,a,pt %xcc, etrap_spill_fixup_32bit; \ ba,a,pt %xcc, etrap_spill_fixup_32bit; \ ba,a,pt %xcc, etrap_spill_fixup_32bit; @@ -592,7 +596,9 @@ user_rtt_fill_64bit: \ /* Normal 32bit fill */ #define FILL_2_GENERIC(ASI) \ - srl %sp, 0, %sp; \ + and %sp, 1, %g3; \ + brnz,pn %g3, (. - (128 + 4)); \ + srl %sp, 0, %sp; \ lduwa [%sp + %g0] ASI, %l0; \ mov 0x04, %g2; \ mov 0x08, %g3; \ @@ -616,14 +622,16 @@ user_rtt_fill_64bit: \ lduwa [%g1 + %g3] ASI, %i6; \ lduwa [%g1 + %g5] ASI, %i7; \ restored; \ - retry; nop; nop; nop; nop; \ + retry; nop; nop; \ b,a,pt %xcc, fill_fixup_dax; \ b,a,pt %xcc, fill_fixup_mna; \ b,a,pt %xcc, fill_fixup; #define FILL_2_GENERIC_RTRAP \ user_rtt_fill_32bit: \ - srl %sp, 0, %sp; \ + and %sp, 1, %g3; \ + brnz,pn %g3, user_rtt_fill_64bit; \ + srl %sp, 0, %sp; \ lduwa [%sp + 0x00] %asi, %l0; \ lduwa [%sp + 0x04] %asi, %l1; \ lduwa [%sp + 0x08] %asi, %l2; \ @@ -643,7 +651,7 @@ user_rtt_fill_32bit: \ ba,pt %xcc, user_rtt_pre_restore; \ restored; \ nop; nop; nop; nop; nop; \ - nop; nop; nop; nop; nop; \ + nop; nop; nop; \ ba,a,pt %xcc, user_rtt_fill_fixup; \ ba,a,pt %xcc, user_rtt_fill_fixup; \ ba,a,pt %xcc, user_rtt_fill_fixup; diff --git a/arch/sparc/include/uapi/asm/unistd.h b/arch/sparc/include/uapi/asm/unistd.h index 8974ef7ae920..bed86a820d09 100644 --- a/arch/sparc/include/uapi/asm/unistd.h +++ b/arch/sparc/include/uapi/asm/unistd.h @@ -405,8 +405,12 @@ #define __NR_setns 337 #define __NR_process_vm_readv 338 #define __NR_process_vm_writev 339 +#define __NR_kern_features 340 -#define NR_syscalls 340 +#define NR_syscalls 341 + +/* Bitmask values returned from kern_features system call. */ +#define KERN_FEATURE_MIXED_MODE_STACK 0x00000001 #ifdef __32bit_syscall_numbers__ /* Sparc 32-bit only has the "setresuid32", "getresuid32" variants, diff --git a/arch/sparc/kernel/perf_event.c b/arch/sparc/kernel/perf_event.c index 885a8af74064..b5c38faa4ead 100644 --- a/arch/sparc/kernel/perf_event.c +++ b/arch/sparc/kernel/perf_event.c @@ -1762,15 +1762,25 @@ static void perf_callchain_user_32(struct perf_callchain_entry *entry, ufp = regs->u_regs[UREG_I6] & 0xffffffffUL; do { - struct sparc_stackf32 *usf, sf; unsigned long pc; - usf = (struct sparc_stackf32 *) ufp; - if (__copy_from_user_inatomic(&sf, usf, sizeof(sf))) - break; + if (thread32_stack_is_64bit(ufp)) { + struct sparc_stackf *usf, sf; - pc = sf.callers_pc; - ufp = (unsigned long)sf.fp; + ufp += STACK_BIAS; + usf = (struct sparc_stackf *) ufp; + if (__copy_from_user_inatomic(&sf, usf, sizeof(sf))) + break; + pc = sf.callers_pc & 0xffffffff; + ufp = ((unsigned long) sf.fp) & 0xffffffff; + } else { + struct sparc_stackf32 *usf, sf; + usf = (struct sparc_stackf32 *) ufp; + if (__copy_from_user_inatomic(&sf, usf, sizeof(sf))) + break; + pc = sf.callers_pc; + ufp = (unsigned long)sf.fp; + } perf_callchain_store(entry, pc); } while (entry->nr < PERF_MAX_STACK_DEPTH); } diff --git a/arch/sparc/kernel/process_64.c b/arch/sparc/kernel/process_64.c index d778248ef3f8..c6e0c2910043 100644 --- a/arch/sparc/kernel/process_64.c +++ b/arch/sparc/kernel/process_64.c @@ -452,13 +452,16 @@ void flush_thread(void) /* It's a bit more tricky when 64-bit tasks are involved... */ static unsigned long clone_stackframe(unsigned long csp, unsigned long psp) { + bool stack_64bit = test_thread_64bit_stack(psp); unsigned long fp, distance, rval; - if (!(test_thread_flag(TIF_32BIT))) { + if (stack_64bit) { csp += STACK_BIAS; psp += STACK_BIAS; __get_user(fp, &(((struct reg_window __user *)psp)->ins[6])); fp += STACK_BIAS; + if (test_thread_flag(TIF_32BIT)) + fp &= 0xffffffff; } else __get_user(fp, &(((struct reg_window32 __user *)psp)->ins[6])); @@ -472,7 +475,7 @@ static unsigned long clone_stackframe(unsigned long csp, unsigned long psp) rval = (csp - distance); if (copy_in_user((void __user *) rval, (void __user *) psp, distance)) rval = 0; - else if (test_thread_flag(TIF_32BIT)) { + else if (!stack_64bit) { if (put_user(((u32)csp), &(((struct reg_window32 __user *)rval)->ins[6]))) rval = 0; @@ -507,18 +510,18 @@ void synchronize_user_stack(void) flush_user_windows(); if ((window = get_thread_wsaved()) != 0) { - int winsize = sizeof(struct reg_window); - int bias = 0; - - if (test_thread_flag(TIF_32BIT)) - winsize = sizeof(struct reg_window32); - else - bias = STACK_BIAS; - window -= 1; do { - unsigned long sp = (t->rwbuf_stkptrs[window] + bias); struct reg_window *rwin = &t->reg_window[window]; + int winsize = sizeof(struct reg_window); + unsigned long sp; + + sp = t->rwbuf_stkptrs[window]; + + if (test_thread_64bit_stack(sp)) + sp += STACK_BIAS; + else + winsize = sizeof(struct reg_window32); if (!copy_to_user((char __user *)sp, rwin, winsize)) { shift_window_buffer(window, get_thread_wsaved() - 1, t); @@ -544,13 +547,6 @@ void fault_in_user_windows(void) { struct thread_info *t = current_thread_info(); unsigned long window; - int winsize = sizeof(struct reg_window); - int bias = 0; - - if (test_thread_flag(TIF_32BIT)) - winsize = sizeof(struct reg_window32); - else - bias = STACK_BIAS; flush_user_windows(); window = get_thread_wsaved(); @@ -558,8 +554,16 @@ void fault_in_user_windows(void) if (likely(window != 0)) { window -= 1; do { - unsigned long sp = (t->rwbuf_stkptrs[window] + bias); struct reg_window *rwin = &t->reg_window[window]; + int winsize = sizeof(struct reg_window); + unsigned long sp; + + sp = t->rwbuf_stkptrs[window]; + + if (test_thread_64bit_stack(sp)) + sp += STACK_BIAS; + else + winsize = sizeof(struct reg_window32); if (unlikely(sp & 0x7UL)) stack_unaligned(sp); diff --git a/arch/sparc/kernel/ptrace_64.c b/arch/sparc/kernel/ptrace_64.c index 484dabac7045..7ff45e4ba681 100644 --- a/arch/sparc/kernel/ptrace_64.c +++ b/arch/sparc/kernel/ptrace_64.c @@ -151,7 +151,7 @@ static int regwindow64_get(struct task_struct *target, { unsigned long rw_addr = regs->u_regs[UREG_I6]; - if (test_tsk_thread_flag(current, TIF_32BIT)) { + if (!test_thread_64bit_stack(rw_addr)) { struct reg_window32 win32; int i; @@ -176,7 +176,7 @@ static int regwindow64_set(struct task_struct *target, { unsigned long rw_addr = regs->u_regs[UREG_I6]; - if (test_tsk_thread_flag(current, TIF_32BIT)) { + if (!test_thread_64bit_stack(rw_addr)) { struct reg_window32 win32; int i; diff --git a/arch/sparc/kernel/sys_sparc_64.c b/arch/sparc/kernel/sys_sparc_64.c index 11c6c9603e71..878ef3d5fec5 100644 --- a/arch/sparc/kernel/sys_sparc_64.c +++ b/arch/sparc/kernel/sys_sparc_64.c @@ -751,3 +751,8 @@ int kernel_execve(const char *filename, : "cc"); return __res; } + +asmlinkage long sys_kern_features(void) +{ + return KERN_FEATURE_MIXED_MODE_STACK; +} diff --git a/arch/sparc/kernel/systbls_64.S b/arch/sparc/kernel/systbls_64.S index 3a58e0d66f51..45ce6be088e4 100644 --- a/arch/sparc/kernel/systbls_64.S +++ b/arch/sparc/kernel/systbls_64.S @@ -86,6 +86,7 @@ sys_call_table32: .word compat_sys_pwritev, compat_sys_rt_tgsigqueueinfo, sys_perf_event_open, compat_sys_recvmmsg, sys_fanotify_init /*330*/ .word sys32_fanotify_mark, sys_prlimit64, sys_name_to_handle_at, compat_sys_open_by_handle_at, compat_sys_clock_adjtime .word sys_syncfs, compat_sys_sendmmsg, sys_setns, compat_sys_process_vm_readv, compat_sys_process_vm_writev +/*340*/ .word sys_kern_features #endif /* CONFIG_COMPAT */ @@ -163,3 +164,4 @@ sys_call_table: .word sys_pwritev, sys_rt_tgsigqueueinfo, sys_perf_event_open, sys_recvmmsg, sys_fanotify_init /*330*/ .word sys_fanotify_mark, sys_prlimit64, sys_name_to_handle_at, sys_open_by_handle_at, sys_clock_adjtime .word sys_syncfs, sys_sendmmsg, sys_setns, sys_process_vm_readv, sys_process_vm_writev +/*340*/ .word sys_kern_features diff --git a/arch/sparc/kernel/unaligned_64.c b/arch/sparc/kernel/unaligned_64.c index f81d038f7340..8201c25e7669 100644 --- a/arch/sparc/kernel/unaligned_64.c +++ b/arch/sparc/kernel/unaligned_64.c @@ -113,21 +113,24 @@ static inline long sign_extend_imm13(long imm) static unsigned long fetch_reg(unsigned int reg, struct pt_regs *regs) { - unsigned long value; + unsigned long value, fp; if (reg < 16) return (!reg ? 0 : regs->u_regs[reg]); + + fp = regs->u_regs[UREG_FP]; + if (regs->tstate & TSTATE_PRIV) { struct reg_window *win; - win = (struct reg_window *)(regs->u_regs[UREG_FP] + STACK_BIAS); + win = (struct reg_window *)(fp + STACK_BIAS); value = win->locals[reg - 16]; - } else if (test_thread_flag(TIF_32BIT)) { + } else if (!test_thread_64bit_stack(fp)) { struct reg_window32 __user *win32; - win32 = (struct reg_window32 __user *)((unsigned long)((u32)regs->u_regs[UREG_FP])); + win32 = (struct reg_window32 __user *)((unsigned long)((u32)fp)); get_user(value, &win32->locals[reg - 16]); } else { struct reg_window __user *win; - win = (struct reg_window __user *)(regs->u_regs[UREG_FP] + STACK_BIAS); + win = (struct reg_window __user *)(fp + STACK_BIAS); get_user(value, &win->locals[reg - 16]); } return value; @@ -135,19 +138,24 @@ static unsigned long fetch_reg(unsigned int reg, struct pt_regs *regs) static unsigned long *fetch_reg_addr(unsigned int reg, struct pt_regs *regs) { + unsigned long fp; + if (reg < 16) return ®s->u_regs[reg]; + + fp = regs->u_regs[UREG_FP]; + if (regs->tstate & TSTATE_PRIV) { struct reg_window *win; - win = (struct reg_window *)(regs->u_regs[UREG_FP] + STACK_BIAS); + win = (struct reg_window *)(fp + STACK_BIAS); return &win->locals[reg - 16]; - } else if (test_thread_flag(TIF_32BIT)) { + } else if (!test_thread_64bit_stack(fp)) { struct reg_window32 *win32; - win32 = (struct reg_window32 *)((unsigned long)((u32)regs->u_regs[UREG_FP])); + win32 = (struct reg_window32 *)((unsigned long)((u32)fp)); return (unsigned long *)&win32->locals[reg - 16]; } else { struct reg_window *win; - win = (struct reg_window *)(regs->u_regs[UREG_FP] + STACK_BIAS); + win = (struct reg_window *)(fp + STACK_BIAS); return &win->locals[reg - 16]; } } @@ -392,13 +400,15 @@ int handle_popc(u32 insn, struct pt_regs *regs) if (rd) regs->u_regs[rd] = ret; } else { - if (test_thread_flag(TIF_32BIT)) { + unsigned long fp = regs->u_regs[UREG_FP]; + + if (!test_thread_64bit_stack(fp)) { struct reg_window32 __user *win32; - win32 = (struct reg_window32 __user *)((unsigned long)((u32)regs->u_regs[UREG_FP])); + win32 = (struct reg_window32 __user *)((unsigned long)((u32)fp)); put_user(ret, &win32->locals[rd - 16]); } else { struct reg_window __user *win; - win = (struct reg_window __user *)(regs->u_regs[UREG_FP] + STACK_BIAS); + win = (struct reg_window __user *)(fp + STACK_BIAS); put_user(ret, &win->locals[rd - 16]); } } @@ -554,7 +564,7 @@ void handle_ld_nf(u32 insn, struct pt_regs *regs) reg[0] = 0; if ((insn & 0x780000) == 0x180000) reg[1] = 0; - } else if (test_thread_flag(TIF_32BIT)) { + } else if (!test_thread_64bit_stack(regs->u_regs[UREG_FP])) { put_user(0, (int __user *) reg); if ((insn & 0x780000) == 0x180000) put_user(0, ((int __user *) reg) + 1); diff --git a/arch/sparc/kernel/visemul.c b/arch/sparc/kernel/visemul.c index 08e074b7eb6a..c096c624ac4d 100644 --- a/arch/sparc/kernel/visemul.c +++ b/arch/sparc/kernel/visemul.c @@ -149,21 +149,24 @@ static inline void maybe_flush_windows(unsigned int rs1, unsigned int rs2, static unsigned long fetch_reg(unsigned int reg, struct pt_regs *regs) { - unsigned long value; + unsigned long value, fp; if (reg < 16) return (!reg ? 0 : regs->u_regs[reg]); + + fp = regs->u_regs[UREG_FP]; + if (regs->tstate & TSTATE_PRIV) { struct reg_window *win; - win = (struct reg_window *)(regs->u_regs[UREG_FP] + STACK_BIAS); + win = (struct reg_window *)(fp + STACK_BIAS); value = win->locals[reg - 16]; - } else if (test_thread_flag(TIF_32BIT)) { + } else if (!test_thread_64bit_stack(fp)) { struct reg_window32 __user *win32; - win32 = (struct reg_window32 __user *)((unsigned long)((u32)regs->u_regs[UREG_FP])); + win32 = (struct reg_window32 __user *)((unsigned long)((u32)fp)); get_user(value, &win32->locals[reg - 16]); } else { struct reg_window __user *win; - win = (struct reg_window __user *)(regs->u_regs[UREG_FP] + STACK_BIAS); + win = (struct reg_window __user *)(fp + STACK_BIAS); get_user(value, &win->locals[reg - 16]); } return value; @@ -172,16 +175,18 @@ static unsigned long fetch_reg(unsigned int reg, struct pt_regs *regs) static inline unsigned long __user *__fetch_reg_addr_user(unsigned int reg, struct pt_regs *regs) { + unsigned long fp = regs->u_regs[UREG_FP]; + BUG_ON(reg < 16); BUG_ON(regs->tstate & TSTATE_PRIV); - if (test_thread_flag(TIF_32BIT)) { + if (!test_thread_64bit_stack(fp)) { struct reg_window32 __user *win32; - win32 = (struct reg_window32 __user *)((unsigned long)((u32)regs->u_regs[UREG_FP])); + win32 = (struct reg_window32 __user *)((unsigned long)((u32)fp)); return (unsigned long __user *)&win32->locals[reg - 16]; } else { struct reg_window __user *win; - win = (struct reg_window __user *)(regs->u_regs[UREG_FP] + STACK_BIAS); + win = (struct reg_window __user *)(fp + STACK_BIAS); return &win->locals[reg - 16]; } } @@ -204,7 +209,7 @@ static void store_reg(struct pt_regs *regs, unsigned long val, unsigned long rd) } else { unsigned long __user *rd_user = __fetch_reg_addr_user(rd, regs); - if (test_thread_flag(TIF_32BIT)) + if (!test_thread_64bit_stack(regs->u_regs[UREG_FP])) __put_user((u32)val, (u32 __user *)rd_user); else __put_user(val, rd_user); diff --git a/arch/sparc/kernel/winfixup.S b/arch/sparc/kernel/winfixup.S index a6b0863c27df..1e67ce958369 100644 --- a/arch/sparc/kernel/winfixup.S +++ b/arch/sparc/kernel/winfixup.S @@ -43,6 +43,8 @@ spill_fixup_mna: spill_fixup_dax: TRAP_LOAD_THREAD_REG(%g6, %g1) ldx [%g6 + TI_FLAGS], %g1 + andcc %sp, 0x1, %g0 + movne %icc, 0, %g1 andcc %g1, _TIF_32BIT, %g0 ldub [%g6 + TI_WSAVED], %g1 sll %g1, 3, %g3 diff --git a/arch/sparc/math-emu/math_64.c b/arch/sparc/math-emu/math_64.c index 1704068da928..034aadbff036 100644 --- a/arch/sparc/math-emu/math_64.c +++ b/arch/sparc/math-emu/math_64.c @@ -320,7 +320,7 @@ int do_mathemu(struct pt_regs *regs, struct fpustate *f, bool illegal_insn_trap) XR = 0; else if (freg < 16) XR = regs->u_regs[freg]; - else if (test_thread_flag(TIF_32BIT)) { + else if (!test_thread_64bit_stack(regs->u_regs[UREG_FP])) { struct reg_window32 __user *win32; flushw_user (); win32 = (struct reg_window32 __user *)((unsigned long)((u32)regs->u_regs[UREG_FP])); -- cgit v1.2.3