This should be done by all target and, since commit 53f6672bcf
("gen-icount: use tcg_ctx.tcg_env instead of cpu_env", 2017-06-30),
is causing the NIOS2 target to hang.
This is because the test for "should I exit to the main loop"
was being done with the correct offset to the icount decrementer,
but using TCG temporary 0 (the frame pointer) rather than the
env pointer.
Cc: qemu-stable@nongnu.org
Cc: Marek Vasut <marex@denx.de>
Reported-by: Thomas Huth <thuth@redhat.com>
Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>
---
target/nios2/translate.c | 1 +
1 file changed, 1 insertion(+)
diff --git a/target/nios2/translate.c b/target/nios2/translate.c
index 6b0961837d..54fbe898df 100644
--- a/target/nios2/translate.c
+++ b/target/nios2/translate.c
@@ -948,6 +948,7 @@ void nios2_tcg_init(void)
int i;
cpu_env = tcg_global_reg_new_ptr(TCG_AREG0, "env");
+ tcg_ctx.tcg_env = cpu_env;
for (i = 0; i < NUM_CORE_REGS; i++) {
cpu_R[i] = tcg_global_mem_new(cpu_env,
--
2.14.2
On 10/10/2017 08:18 AM, Paolo Bonzini wrote: > This should be done by all target and, since commit 53f6672bcf > ("gen-icount: use tcg_ctx.tcg_env instead of cpu_env", 2017-06-30), > is causing the NIOS2 target to hang. > > This is because the test for "should I exit to the main loop" > was being done with the correct offset to the icount decrementer, > but using TCG temporary 0 (the frame pointer) rather than the > env pointer. > > Cc: qemu-stable@nongnu.org > Cc: Marek Vasut <marex@denx.de> > Reported-by: Thomas Huth <thuth@redhat.com> > Signed-off-by: Paolo Bonzini <pbonzini@redhat.com> > --- > target/nios2/translate.c | 1 + > 1 file changed, 1 insertion(+) Reviewed-by: Richard Henderson <richard.henderson@linaro.org> r~
On 10 October 2017 at 16:18, Paolo Bonzini <pbonzini@redhat.com> wrote: > This should be done by all target and, since commit 53f6672bcf > ("gen-icount: use tcg_ctx.tcg_env instead of cpu_env", 2017-06-30), > is causing the NIOS2 target to hang. > > This is because the test for "should I exit to the main loop" > was being done with the correct offset to the icount decrementer, > but using TCG temporary 0 (the frame pointer) rather than the > env pointer. > > Cc: qemu-stable@nongnu.org > Cc: Marek Vasut <marex@denx.de> > Reported-by: Thomas Huth <thuth@redhat.com> > Signed-off-by: Paolo Bonzini <pbonzini@redhat.com> > --- > target/nios2/translate.c | 1 + > 1 file changed, 1 insertion(+) > > diff --git a/target/nios2/translate.c b/target/nios2/translate.c > index 6b0961837d..54fbe898df 100644 > --- a/target/nios2/translate.c > +++ b/target/nios2/translate.c > @@ -948,6 +948,7 @@ void nios2_tcg_init(void) > int i; > > cpu_env = tcg_global_reg_new_ptr(TCG_AREG0, "env"); > + tcg_ctx.tcg_env = cpu_env; > > for (i = 0; i < NUM_CORE_REGS; i++) { > cpu_R[i] = tcg_global_mem_new(cpu_env, > -- Is there somewhere we could put an assert() so we can catch this error for future targets rather than them just silently generating bad code? thanks -- PMM
On 10/10/2017 09:49 AM, Peter Maydell wrote: > On 10 October 2017 at 16:18, Paolo Bonzini <pbonzini@redhat.com> wrote: >> This should be done by all target and, since commit 53f6672bcf >> ("gen-icount: use tcg_ctx.tcg_env instead of cpu_env", 2017-06-30), >> is causing the NIOS2 target to hang. >> >> This is because the test for "should I exit to the main loop" >> was being done with the correct offset to the icount decrementer, >> but using TCG temporary 0 (the frame pointer) rather than the >> env pointer. >> >> Cc: qemu-stable@nongnu.org >> Cc: Marek Vasut <marex@denx.de> >> Reported-by: Thomas Huth <thuth@redhat.com> >> Signed-off-by: Paolo Bonzini <pbonzini@redhat.com> >> --- >> target/nios2/translate.c | 1 + >> 1 file changed, 1 insertion(+) >> >> diff --git a/target/nios2/translate.c b/target/nios2/translate.c >> index 6b0961837d..54fbe898df 100644 >> --- a/target/nios2/translate.c >> +++ b/target/nios2/translate.c >> @@ -948,6 +948,7 @@ void nios2_tcg_init(void) >> int i; >> >> cpu_env = tcg_global_reg_new_ptr(TCG_AREG0, "env"); >> + tcg_ctx.tcg_env = cpu_env; >> >> for (i = 0; i < NUM_CORE_REGS; i++) { >> cpu_R[i] = tcg_global_mem_new(cpu_env, >> -- > > Is there somewhere we could put an assert() so we can catch this > error for future targets rather than them just silently generating > bad code? Yes, I can arrange for this. That said, probably the best change we could make is to eliminate cpu_env entirely, and create tcg_ctx.tcg_env generically (possibly renaming this to a standalone global variable called cpu_env, minimizing the changes within target/*). It is, after all, the same for every target, and there is no longer any other TCG_AREG* value that can be used. r~
© 2016 - 2024 Red Hat, Inc.