MIPS: Basic MSA context switching support

This patch adds support for context switching the MSA vector registers.
These 128 bit vector registers are aliased with the FP registers - an
FP register accesses the least significant bits of the vector register
with which it is aliased (ie. the register with the same index). Due to
both this & the requirement that the scalar FPU must be 64-bit (FR=1) if
enabled at the same time as MSA the kernel will enable MSA & scalar FP
at the same time for tasks which use MSA. If we restore the MSA vector
context then we might as well enable the scalar FPU since the reason it
was left disabled was to allow for lazy FP context restoring - but we
just restored the FP context as it's a subset of the vector context. If
we restore the FP context and have previously used MSA then we have to
restore the whole vector context anyway (see comment in
enable_restore_fp_context for details) so similarly we might as well
enable MSA.

Thus if a task does not use MSA then it will continue to behave as
without this patch - the scalar FP context will be saved & restored as
usual. But if a task executes an MSA instruction then it will save &
restore the vector context forever more.

Signed-off-by: Paul Burton <paul.burton@imgtec.com>
Cc: linux-mips@linux-mips.org
Patchwork: https://patchwork.linux-mips.org/patch/6431/
Signed-off-by: Ralf Baechle <ralf@linux-mips.org>

4 files changed, 143 insertions, 24 deletions

diff --git a/arch/mips/kernel/genex.S b/arch/mips/kernel/genex.S
index d84f6a509502..278a49bd93b6 100644
--- a/arch/mips/kernel/genex.S
+++ b/arch/mips/kernel/genex.S
@@ -477,6 +477,7 @@ NESTED(nmi_handler, PT_SIZE, sp)
 	BUILD_HANDLER tr tr sti silent			/* #13 */
 	BUILD_HANDLER fpe fpe fpe silent		/* #15 */
 	BUILD_HANDLER ftlb ftlb none silent		/* #16 */
+	BUILD_HANDLER msa msa sti silent		/* #21 */
 	BUILD_HANDLER mdmx mdmx sti silent		/* #22 */
 #ifdef	CONFIG_HARDWARE_WATCHPOINTS
 	/*
diff --git a/arch/mips/kernel/process.c b/arch/mips/kernel/process.c
index 2f01f3d48d75..60e39dc7f1eb 100644
--- a/arch/mips/kernel/process.c
+++ b/arch/mips/kernel/process.c
@@ -32,6 +32,7 @@
 #include <asm/cpu.h>
 #include <asm/dsp.h>
 #include <asm/fpu.h>
+#include <asm/msa.h>
 #include <asm/pgtable.h>
 #include <asm/mipsregs.h>
 #include <asm/processor.h>
@@ -65,6 +66,8 @@ void start_thread(struct pt_regs * regs, unsigned long pc, unsigned long sp)
 	clear_used_math();
 	clear_fpu_owner();
 	init_dsp();
+	clear_thread_flag(TIF_MSA_CTX_LIVE);
+	disable_msa();
 	regs->cp0_epc = pc;
 	regs->regs[29] = sp;
 }
@@ -89,7 +92,9 @@ int copy_thread(unsigned long clone_flags, unsigned long usp,
 
 	preempt_disable();
 
-	if (is_fpu_owner())
+	if (is_msa_enabled())
+		save_msa(p);
+	else if (is_fpu_owner())
 		save_fp(p);
 
 	if (cpu_has_dsp)
diff --git a/arch/mips/kernel/r4k_switch.S b/arch/mips/kernel/r4k_switch.S
index cc78dd9a17c7..f938ecd22af0 100644
--- a/arch/mips/kernel/r4k_switch.S
+++ b/arch/mips/kernel/r4k_switch.S
@@ -29,18 +29,8 @@
 #define ST_OFF (_THREAD_SIZE - 32 - PT_SIZE + PT_STATUS)
 
 /*
- * FPU context is saved iff the process has used it's FPU in the current
- * time slice as indicated by _TIF_USEDFPU.  In any case, the CU1 bit for user
- * space STATUS register should be 0, so that a process *always* starts its
- * userland with FPU disabled after each context switch.
- *
- * FPU will be enabled as soon as the process accesses FPU again, through
- * do_cpu() trap.
- */
-
-/*
  * task_struct *resume(task_struct *prev, task_struct *next,
- *		       struct thread_info *next_ti, int usedfpu)
+ *		       struct thread_info *next_ti, s32 fp_save)
  */
 	.align	5
 	LEAF(resume)
@@ -50,23 +40,37 @@
 	LONG_S	ra, THREAD_REG31(a0)
 
 	/*
-	 * check if we need to save FPU registers
+	 * Check whether we need to save any FP context. FP context is saved
+	 * iff the process has used the context with the scalar FPU or the MSA
+	 * ASE in the current time slice, as indicated by _TIF_USEDFPU and
+	 * _TIF_USEDMSA respectively. switch_to will have set fp_save
+	 * accordingly to an FP_SAVE_ enum value.
 	 */
+	beqz	a3, 2f
 
-	beqz	a3, 1f
-
-	PTR_L	t3, TASK_THREAD_INFO(a0)
 	/*
-	 * clear saved user stack CU1 bit
+	 * We do. Clear the saved CU1 bit for prev, such that next time it is
+	 * scheduled it will start in userland with the FPU disabled. If the
+	 * task uses the FPU then it will be enabled again via the do_cpu trap.
+	 * This allows us to lazily restore the FP context.
 	 */
+	PTR_L	t3, TASK_THREAD_INFO(a0)
 	LONG_L	t0, ST_OFF(t3)
 	li	t1, ~ST0_CU1
 	and	t0, t0, t1
 	LONG_S	t0, ST_OFF(t3)
 
+	/* Check whether we're saving scalar or vector context. */
+	bgtz	a3, 1f
+
+	/* Save 128b MSA vector context. */
+	msa_save_all	a0
+	b	2f
+
+1:	/* Save 32b/64b scalar FP context. */
 	fpu_save_double a0 t0 t1		# c0_status passed in t0
 						# clobbers t1
-1:
+2:
 
 #if defined(CONFIG_CC_STACKPROTECTOR) && !defined(CONFIG_SMP)
 	PTR_LA	t8, __stack_chk_guard
@@ -141,6 +145,26 @@ LEAF(_restore_fp)
 	jr	ra
 	END(_restore_fp)
 
+#ifdef CONFIG_CPU_HAS_MSA
+
+/*
+ * Save a thread's MSA vector context.
+ */
+LEAF(_save_msa)
+	msa_save_all	a0
+	jr	ra
+	END(_save_msa)
+
+/*
+ * Restore a thread's MSA vector context.
+ */
+LEAF(_restore_msa)
+	msa_restore_all	a0
+	jr	ra
+	END(_restore_msa)
+
+#endif
+
 /*
  * Load the FPU with signalling NANS.  This bit pattern we're using has
  * the property that no matter whether considered as single or as double
diff --git a/arch/mips/kernel/traps.c b/arch/mips/kernel/traps.c
index b0c7f80e05e5..4f27cbfad1e5 100644
--- a/arch/mips/kernel/traps.c
+++ b/arch/mips/kernel/traps.c
@@ -47,6 +47,7 @@
 #include <asm/mipsregs.h>
 #include <asm/mipsmtregs.h>
 #include <asm/module.h>
+#include <asm/msa.h>
 #include <asm/pgtable.h>
 #include <asm/ptrace.h>
 #include <asm/sections.h>
@@ -79,6 +80,7 @@ extern asmlinkage void handle_ov(void);
 extern asmlinkage void handle_tr(void);
 extern asmlinkage void handle_fpe(void);
 extern asmlinkage void handle_ftlb(void);
+extern asmlinkage void handle_msa(void);
 extern asmlinkage void handle_mdmx(void);
 extern asmlinkage void handle_watch(void);
 extern asmlinkage void handle_mt(void);
@@ -1074,6 +1076,76 @@ static int default_cu2_call(struct notifier_block *nfb, unsigned long action,
 	return NOTIFY_OK;
 }
 
+static int enable_restore_fp_context(int msa)
+{
+	int err, was_fpu_owner;
+
+	if (!used_math()) {
+		/* First time FP context user. */
+		err = init_fpu();
+		if (msa && !err)
+			enable_msa();
+		if (!err)
+			set_used_math();
+		return err;
+	}
+
+	/*
+	 * This task has formerly used the FP context.
+	 *
+	 * If this thread has no live MSA vector context then we can simply
+	 * restore the scalar FP context. If it has live MSA vector context
+	 * (that is, it has or may have used MSA since last performing a
+	 * function call) then we'll need to restore the vector context. This
+	 * applies even if we're currently only executing a scalar FP
+	 * instruction. This is because if we were to later execute an MSA
+	 * instruction then we'd either have to:
+	 *
+	 *  - Restore the vector context & clobber any registers modified by
+	 *    scalar FP instructions between now & then.
+	 *
+	 * or
+	 *
+	 *  - Not restore the vector context & lose the most significant bits
+	 *    of all vector registers.
+	 *
+	 * Neither of those options is acceptable. We cannot restore the least
+	 * significant bits of the registers now & only restore the most
+	 * significant bits later because the most significant bits of any
+	 * vector registers whose aliased FP register is modified now will have
+	 * been zeroed. We'd have no way to know that when restoring the vector
+	 * context & thus may load an outdated value for the most significant
+	 * bits of a vector register.
+	 */
+	if (!msa && !thread_msa_context_live())
+		return own_fpu(1);
+
+	/*
+	 * This task is using or has previously used MSA. Thus we require
+	 * that Status.FR == 1.
+	 */
+	was_fpu_owner = is_fpu_owner();
+	err = own_fpu(0);
+	if (err)
+		return err;
+
+	enable_msa();
+	write_msa_csr(current->thread.fpu.msacsr);
+	set_thread_flag(TIF_USEDMSA);
+
+	/*
+	 * If this is the first time that the task is using MSA and it has
+	 * previously used scalar FP in this time slice then we already nave
+	 * FP context which we shouldn't clobber.
+	 */
+	if (!test_and_set_thread_flag(TIF_MSA_CTX_LIVE) && was_fpu_owner)
+		return 0;
+
+	/* We need to restore the vector context. */
+	restore_msa(current);
+	return 0;
+}
+
 asmlinkage void do_cpu(struct pt_regs *regs)
 {
 	enum ctx_state prev_state;
@@ -1153,12 +1225,7 @@ asmlinkage void do_cpu(struct pt_regs *regs)
 		/* Fall through.  */
 
 	case 1:
-		if (used_math())	/* Using the FPU again.	 */
-			err = own_fpu(1);
-		else {			/* First time FPU user.	 */
-			err = init_fpu();
-			set_used_math();
-		}
+		err = enable_restore_fp_context(0);
 
 		if (!raw_cpu_has_fpu || err) {
 			int sig;
@@ -1183,6 +1250,27 @@ out:
 	exception_exit(prev_state);
 }
 
+asmlinkage void do_msa(struct pt_regs *regs)
+{
+	enum ctx_state prev_state;
+	int err;
+
+	prev_state = exception_enter();
+
+	if (!cpu_has_msa || test_thread_flag(TIF_32BIT_FPREGS)) {
+		force_sig(SIGILL, current);
+		goto out;
+	}
+
+	die_if_kernel("do_msa invoked from kernel context!", regs);
+
+	err = enable_restore_fp_context(1);
+	if (err)
+		force_sig(SIGILL, current);
+out:
+	exception_exit(prev_state);
+}
+
 asmlinkage void do_mdmx(struct pt_regs *regs)
 {
 	enum ctx_state prev_state;
@@ -2041,6 +2129,7 @@ void __init trap_init(void)
 		set_except_vector(15, handle_fpe);
 
 	set_except_vector(16, handle_ftlb);
+	set_except_vector(21, handle_msa);
 	set_except_vector(22, handle_mdmx);
 
 	if (cpu_has_mcheck)