aboutsummaryrefslogtreecommitdiffstats
path: root/arch/powerpc/kernel
diff options
context:
space:
mode:
authorDavid Gibson <david@gibson.dropbear.id.au>2005-10-27 02:27:25 -0400
committerPaul Mackerras <paulus@samba.org>2005-10-27 06:48:50 -0400
commit25c8a78b1e00ac0cc640677eda78b462c2cd4c6e (patch)
treea0044f8b2b557799a8cb3346b590fcd3a8507ed7 /arch/powerpc/kernel
parentfda262b8978d0089758ef9444508434c74113a61 (diff)
[PATCH] powerpc: Fix handling of fpscr on 64-bit
The recent merge of fpu.S broken the handling of fpscr for ARCH=powerpc and CONFIG_PPC64=y. FP registers could be corrupted, leading to strange random application crashes. The confusion arises, because the thread_struct has (and requires) a 64-bit area to save the fpscr, because we use load/store double instructions to get it in to/out of the FPU. However, only the low 32-bits are actually used, so we want to treat it as a 32-bit quantity when manipulating its bits to avoid extra load/stores on 32-bit. This patch replaces the current definition with a structure of two 32-bit quantities (pad and val), to clarify things as much as is possible. The 'val' field is used when manipulating bits, the structure itself is used when obtaining the address for loading/unloading the value from the FPU. While we're at it, consolidate the 4 (!) almost identical versions of cvt_fd() and cvt_df() (arch/ppc/kernel/misc.S, arch/ppc64/kernel/misc.S, arch/powerpc/kernel/misc_32.S, arch/powerpc/kernel/misc_64.S) into a single version in fpu.S. The new version takes a pointer to thread_struct and applies the correct offset itself, rather than a pointer to the fpscr field itself, again to avoid confusion as to which is the correct field to use. Finally, this patch makes ARCH=ppc64 also use the consolidated fpu.S code, which it previously did not. Built for G5 (ARCH=ppc64 and ARCH=powerpc), 32-bit powermac (ARCH=ppc and ARCH=powerpc) and Walnut (ARCH=ppc, CONFIG_MATH_EMULATION=y). Booted on G5 (ARCH=powerpc) and things which previously fell over no longer do. Signed-off-by: David Gibson <dwg@au1.ibm.com> Signed-off-by: Paul Mackerras <paulus@samba.org>
Diffstat (limited to 'arch/powerpc/kernel')
-rw-r--r--arch/powerpc/kernel/Makefile5
-rw-r--r--arch/powerpc/kernel/fpu.S31
-rw-r--r--arch/powerpc/kernel/misc_32.S27
-rw-r--r--arch/powerpc/kernel/misc_64.S19
-rw-r--r--arch/powerpc/kernel/process.c2
-rw-r--r--arch/powerpc/kernel/signal_32.c2
-rw-r--r--arch/powerpc/kernel/traps.c2
7 files changed, 33 insertions, 55 deletions
diff --git a/arch/powerpc/kernel/Makefile b/arch/powerpc/kernel/Makefile
index a733347964a0..94cf917b7854 100644
--- a/arch/powerpc/kernel/Makefile
+++ b/arch/powerpc/kernel/Makefile
@@ -29,7 +29,6 @@ extra-$(CONFIG_44x) := head_44x.o
29extra-$(CONFIG_FSL_BOOKE) := head_fsl_booke.o 29extra-$(CONFIG_FSL_BOOKE) := head_fsl_booke.o
30extra-$(CONFIG_8xx) := head_8xx.o 30extra-$(CONFIG_8xx) := head_8xx.o
31extra-$(CONFIG_PPC64) += entry_64.o 31extra-$(CONFIG_PPC64) += entry_64.o
32extra-$(CONFIG_PPC_FPU) += fpu.o
33extra-y += vmlinux.lds 32extra-y += vmlinux.lds
34 33
35obj-y += process.o init_task.o time.o \ 34obj-y += process.o init_task.o time.o \
@@ -49,7 +48,7 @@ else
49# stuff used from here for ARCH=ppc or ARCH=ppc64 48# stuff used from here for ARCH=ppc or ARCH=ppc64
50obj-$(CONFIG_PPC64) += traps.o process.o init_task.o time.o 49obj-$(CONFIG_PPC64) += traps.o process.o init_task.o time.o
51 50
52fpux-$(CONFIG_PPC32) += fpu.o
53extra-$(CONFIG_PPC_FPU) += $(fpux-y)
54 51
55endif 52endif
53
54extra-$(CONFIG_PPC_FPU) += fpu.o
diff --git a/arch/powerpc/kernel/fpu.S b/arch/powerpc/kernel/fpu.S
index 563d445ff584..51fd78da25b7 100644
--- a/arch/powerpc/kernel/fpu.S
+++ b/arch/powerpc/kernel/fpu.S
@@ -48,7 +48,7 @@ _GLOBAL(load_up_fpu)
48 addi r4,r4,THREAD /* want last_task_used_math->thread */ 48 addi r4,r4,THREAD /* want last_task_used_math->thread */
49 SAVE_32FPRS(0, r4) 49 SAVE_32FPRS(0, r4)
50 mffs fr0 50 mffs fr0
51 stfd fr0,THREAD_FPSCR-4(r4) 51 stfd fr0,THREAD_FPSCR(r4)
52 LDL r5,PT_REGS(r4) 52 LDL r5,PT_REGS(r4)
53 tophys(r5,r5) 53 tophys(r5,r5)
54 LDL r4,_MSR-STACK_FRAME_OVERHEAD(r5) 54 LDL r4,_MSR-STACK_FRAME_OVERHEAD(r5)
@@ -71,7 +71,7 @@ _GLOBAL(load_up_fpu)
71 or r12,r12,r4 71 or r12,r12,r4
72 std r12,_MSR(r1) 72 std r12,_MSR(r1)
73#endif 73#endif
74 lfd fr0,THREAD_FPSCR-4(r5) 74 lfd fr0,THREAD_FPSCR(r5)
75 mtfsf 0xff,fr0 75 mtfsf 0xff,fr0
76 REST_32FPRS(0, r5) 76 REST_32FPRS(0, r5)
77#ifndef CONFIG_SMP 77#ifndef CONFIG_SMP
@@ -104,7 +104,7 @@ _GLOBAL(giveup_fpu)
104 CMPI 0,r5,0 104 CMPI 0,r5,0
105 SAVE_32FPRS(0, r3) 105 SAVE_32FPRS(0, r3)
106 mffs fr0 106 mffs fr0
107 stfd fr0,THREAD_FPSCR-4(r3) 107 stfd fr0,THREAD_FPSCR(r3)
108 beq 1f 108 beq 1f
109 LDL r4,_MSR-STACK_FRAME_OVERHEAD(r5) 109 LDL r4,_MSR-STACK_FRAME_OVERHEAD(r5)
110 li r3,MSR_FP|MSR_FE0|MSR_FE1 110 li r3,MSR_FP|MSR_FE0|MSR_FE1
@@ -117,3 +117,28 @@ _GLOBAL(giveup_fpu)
117 STL r5,OFF(last_task_used_math)(r4) 117 STL r5,OFF(last_task_used_math)(r4)
118#endif /* CONFIG_SMP */ 118#endif /* CONFIG_SMP */
119 blr 119 blr
120
121/*
122 * These are used in the alignment trap handler when emulating
123 * single-precision loads and stores.
124 * We restore and save the fpscr so the task gets the same result
125 * and exceptions as if the cpu had performed the load or store.
126 */
127
128_GLOBAL(cvt_fd)
129 lfd 0,THREAD_FPSCR(r5) /* load up fpscr value */
130 mtfsf 0xff,0
131 lfs 0,0(r3)
132 stfd 0,0(r4)
133 mffs 0
134 stfd 0,THREAD_FPSCR(r5) /* save new fpscr value */
135 blr
136
137_GLOBAL(cvt_df)
138 lfd 0,THREAD_FPSCR(r5) /* load up fpscr value */
139 mtfsf 0xff,0
140 lfd 0,0(r3)
141 stfs 0,0(r4)
142 mffs 0
143 stfd 0,THREAD_FPSCR(r5) /* save new fpscr value */
144 blr
diff --git a/arch/powerpc/kernel/misc_32.S b/arch/powerpc/kernel/misc_32.S
index 303229b090b8..3bedb532aed9 100644
--- a/arch/powerpc/kernel/misc_32.S
+++ b/arch/powerpc/kernel/misc_32.S
@@ -993,33 +993,6 @@ _GLOBAL(_get_SP)
993 blr 993 blr
994 994
995/* 995/*
996 * These are used in the alignment trap handler when emulating
997 * single-precision loads and stores.
998 * We restore and save the fpscr so the task gets the same result
999 * and exceptions as if the cpu had performed the load or store.
1000 */
1001
1002#ifdef CONFIG_PPC_FPU
1003_GLOBAL(cvt_fd)
1004 lfd 0,-4(r5) /* load up fpscr value */
1005 mtfsf 0xff,0
1006 lfs 0,0(r3)
1007 stfd 0,0(r4)
1008 mffs 0 /* save new fpscr value */
1009 stfd 0,-4(r5)
1010 blr
1011
1012_GLOBAL(cvt_df)
1013 lfd 0,-4(r5) /* load up fpscr value */
1014 mtfsf 0xff,0
1015 lfd 0,0(r3)
1016 stfs 0,0(r4)
1017 mffs 0 /* save new fpscr value */
1018 stfd 0,-4(r5)
1019 blr
1020#endif
1021
1022/*
1023 * Create a kernel thread 996 * Create a kernel thread
1024 * kernel_thread(fn, arg, flags) 997 * kernel_thread(fn, arg, flags)
1025 */ 998 */
diff --git a/arch/powerpc/kernel/misc_64.S b/arch/powerpc/kernel/misc_64.S
index 4775bed42cac..b3e95ff0dba0 100644
--- a/arch/powerpc/kernel/misc_64.S
+++ b/arch/powerpc/kernel/misc_64.S
@@ -462,25 +462,6 @@ _GLOBAL(_outsl_ns)
462 sync 462 sync
463 blr 463 blr
464 464
465
466_GLOBAL(cvt_fd)
467 lfd 0,0(r5) /* load up fpscr value */
468 mtfsf 0xff,0
469 lfs 0,0(r3)
470 stfd 0,0(r4)
471 mffs 0 /* save new fpscr value */
472 stfd 0,0(r5)
473 blr
474
475_GLOBAL(cvt_df)
476 lfd 0,0(r5) /* load up fpscr value */
477 mtfsf 0xff,0
478 lfd 0,0(r3)
479 stfs 0,0(r4)
480 mffs 0 /* save new fpscr value */
481 stfd 0,0(r5)
482 blr
483
484/* 465/*
485 * identify_cpu and calls setup_cpu 466 * identify_cpu and calls setup_cpu
486 * In: r3 = base of the cpu_specs array 467 * In: r3 = base of the cpu_specs array
diff --git a/arch/powerpc/kernel/process.c b/arch/powerpc/kernel/process.c
index 047da1ae21fe..8f85dabe4df3 100644
--- a/arch/powerpc/kernel/process.c
+++ b/arch/powerpc/kernel/process.c
@@ -665,7 +665,7 @@ void start_thread(struct pt_regs *regs, unsigned long start, unsigned long sp)
665#endif 665#endif
666#endif /* CONFIG_SMP */ 666#endif /* CONFIG_SMP */
667 memset(current->thread.fpr, 0, sizeof(current->thread.fpr)); 667 memset(current->thread.fpr, 0, sizeof(current->thread.fpr));
668 current->thread.fpscr = 0; 668 current->thread.fpscr.val = 0;
669#ifdef CONFIG_ALTIVEC 669#ifdef CONFIG_ALTIVEC
670 memset(current->thread.vr, 0, sizeof(current->thread.vr)); 670 memset(current->thread.vr, 0, sizeof(current->thread.vr));
671 memset(&current->thread.vscr, 0, sizeof(current->thread.vscr)); 671 memset(&current->thread.vscr, 0, sizeof(current->thread.vscr));
diff --git a/arch/powerpc/kernel/signal_32.c b/arch/powerpc/kernel/signal_32.c
index 92452b2db26a..444c3e81884c 100644
--- a/arch/powerpc/kernel/signal_32.c
+++ b/arch/powerpc/kernel/signal_32.c
@@ -403,7 +403,7 @@ static int save_user_regs(struct pt_regs *regs, struct mcontext __user *frame,
403 ELF_NFPREG * sizeof(double))) 403 ELF_NFPREG * sizeof(double)))
404 return 1; 404 return 1;
405 405
406 current->thread.fpscr = 0; /* turn off all fp exceptions */ 406 current->thread.fpscr.val = 0; /* turn off all fp exceptions */
407 407
408#ifdef CONFIG_ALTIVEC 408#ifdef CONFIG_ALTIVEC
409 /* save altivec registers */ 409 /* save altivec registers */
diff --git a/arch/powerpc/kernel/traps.c b/arch/powerpc/kernel/traps.c
index f87580382da4..5d638ecddbd0 100644
--- a/arch/powerpc/kernel/traps.c
+++ b/arch/powerpc/kernel/traps.c
@@ -549,7 +549,7 @@ static void parse_fpe(struct pt_regs *regs)
549 549
550 flush_fp_to_thread(current); 550 flush_fp_to_thread(current);
551 551
552 fpscr = current->thread.fpscr; 552 fpscr = current->thread.fpscr.val;
553 553
554 /* Invalid operation */ 554 /* Invalid operation */
555 if ((fpscr & FPSCR_VE) && (fpscr & FPSCR_VX)) 555 if ((fpscr & FPSCR_VE) && (fpscr & FPSCR_VX))