[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>

5 files changed, 5 insertions, 32 deletions

diff --git a/arch/ppc/kernel/align.c b/arch/ppc/kernel/align.c
index ff81da9598d..ab398c4b70b 100644
--- a/arch/ppc/kernel/align.c
+++ b/arch/ppc/kernel/align.c
@@ -375,7 +375,7 @@ fix_alignment(struct pt_regs *regs)
 #ifdef CONFIG_PPC_FPU
 		preempt_disable();
 		enable_kernel_fp();
-		cvt_fd(&data.f, &data.d, &current->thread.fpscr);
+		cvt_fd(&data.f, &data.d, &current->thread);
 		preempt_enable();
 #else
 		return 0;
@@ -385,7 +385,7 @@ fix_alignment(struct pt_regs *regs)
 #ifdef CONFIG_PPC_FPU
 		preempt_disable();
 		enable_kernel_fp();
-		cvt_df(&data.d, &data.f, &current->thread.fpscr);
+		cvt_df(&data.d, &data.f, &current->thread);
 		preempt_enable();
 #else
 		return 0;
diff --git a/arch/ppc/kernel/misc.S b/arch/ppc/kernel/misc.S
index 2350f3e09f9..3056ede2424 100644
--- a/arch/ppc/kernel/misc.S
+++ b/arch/ppc/kernel/misc.S
@@ -968,33 +968,6 @@ _GLOBAL(_get_SP)
 	blr
 
 /*
- * These are used in the alignment trap handler when emulating
- * single-precision loads and stores.
- * We restore and save the fpscr so the task gets the same result
- * and exceptions as if the cpu had performed the load or store.
- */
-
-#ifdef CONFIG_PPC_FPU
-_GLOBAL(cvt_fd)
-	lfd	0,-4(r5)	/* load up fpscr value */
-	mtfsf	0xff,0
-	lfs	0,0(r3)
-	stfd	0,0(r4)
-	mffs	0		/* save new fpscr value */
-	stfd	0,-4(r5)
-	blr
-
-_GLOBAL(cvt_df)
-	lfd	0,-4(r5)	/* load up fpscr value */
-	mtfsf	0xff,0
-	lfd	0,0(r3)
-	stfs	0,0(r4)
-	mffs	0		/* save new fpscr value */
-	stfd	0,-4(r5)
-	blr
-#endif
-
-/*
  * Create a kernel thread
  *   kernel_thread(fn, arg, flags)
  */
diff --git a/arch/ppc/kernel/process.c b/arch/ppc/kernel/process.c
index 6d60c40598e..78ea10197a0 100644
--- a/arch/ppc/kernel/process.c
+++ b/arch/ppc/kernel/process.c
@@ -542,7 +542,7 @@ void start_thread(struct pt_regs *regs, unsigned long nip, unsigned long sp)
 		last_task_used_spe = NULL;
 #endif
 	memset(current->thread.fpr, 0, sizeof(current->thread.fpr));
-	current->thread.fpscr = 0;
+	current->thread.fpscr.val = 0;
 #ifdef CONFIG_ALTIVEC
 	memset(current->thread.vr, 0, sizeof(current->thread.vr));
 	memset(&current->thread.vscr, 0, sizeof(current->thread.vscr));
diff --git a/arch/ppc/kernel/traps.c b/arch/ppc/kernel/traps.c
index 5e4bf88a1ef..f265b81e700 100644
--- a/arch/ppc/kernel/traps.c
+++ b/arch/ppc/kernel/traps.c
@@ -659,7 +659,7 @@ void program_check_exception(struct pt_regs *regs)
 			giveup_fpu(current);
 		preempt_enable();
 
-		fpscr = current->thread.fpscr;
+		fpscr = current->thread.fpscr.val;
 		fpscr &= fpscr << 22;	/* mask summary bits with enables */
 		if (fpscr & FPSCR_VX)
 			code = FPE_FLTINV;
diff --git a/arch/ppc/math-emu/sfp-machine.h b/arch/ppc/math-emu/sfp-machine.h
index 686e06d2918..4b17d83cfcd 100644
--- a/arch/ppc/math-emu/sfp-machine.h
+++ b/arch/ppc/math-emu/sfp-machine.h
@@ -166,7 +166,7 @@ extern int  fp_pack_ds(void *, long, unsigned long, unsigned long, long, long);
 #include <linux/kernel.h>
 #include <linux/sched.h>
 
-#define __FPU_FPSCR	(current->thread.fpscr)
+#define __FPU_FPSCR	(current->thread.fpscr.val)
 
 /* We only actually write to the destination register
  * if exceptions signalled (if any) will not trap.