[Qemu-devel] [PATCH] accel/tcg/cputlb: avoid recursive BQL (fixes #1706296)

[Qemu-devel] [PATCH] accel/tcg/cputlb: avoid recursive BQL (fixes #1706296)

[Alex Bennée]

The mmio path (see exec.c:prepare_mmio_access) already protects itself
against recursive locking and it makes sense to do the same for
io_readx/writex. Otherwise any helper running in the BQL context will
assert when it attempts to write to device memory as in the case of
the bug report.

Signed-off-by: Alex Bennée <alex.bennee@linaro.org>
CC: Richard Jones <rjones@redhat.com>
CC: Paolo Bonzini <bonzini@gnu.org>
---
 accel/tcg/cputlb.c | 4 ++--
 1 file changed, 2 insertions(+), 2 deletions(-)

diff --git a/accel/tcg/cputlb.c b/accel/tcg/cputlb.c
index e72415a882..bcbcc4db6c 100644
--- a/accel/tcg/cputlb.c
+++ b/accel/tcg/cputlb.c
@@ -765,7 +765,7 @@ static uint64_t io_readx(CPUArchState *env, CPUIOTLBEntry *iotlbentry,
 
     cpu->mem_io_vaddr = addr;
 
-    if (mr->global_locking) {
+    if (mr->global_locking && !qemu_mutex_iothread_locked()) {
         qemu_mutex_lock_iothread();
         locked = true;
     }
@@ -800,7 +800,7 @@ static void io_writex(CPUArchState *env, CPUIOTLBEntry *iotlbentry,
     cpu->mem_io_vaddr = addr;
     cpu->mem_io_pc = retaddr;
 
-    if (mr->global_locking) {
+    if (mr->global_locking && !qemu_mutex_iothread_locked()) {
         qemu_mutex_lock_iothread();
         locked = true;
     }
-- 
2.14.1

Re: [Qemu-devel] [PATCH] accel/tcg/cputlb: avoid recursive BQL (fixes #1706296)

[Peter Maydell]

On 21 September 2017 at 12:06, Alex Bennée <alex.bennee@linaro.org> wrote:
> The mmio path (see exec.c:prepare_mmio_access) already protects itself
> against recursive locking and it makes sense to do the same for
> io_readx/writex. Otherwise any helper running in the BQL context will
> assert when it attempts to write to device memory as in the case of
> the bug report.
>
> Signed-off-by: Alex Bennée <alex.bennee@linaro.org>
> CC: Richard Jones <rjones@redhat.com>
> CC: Paolo Bonzini <bonzini@gnu.org>

Seems worth:
Cc: qemu-stable@nongnu.org

> ---
>  accel/tcg/cputlb.c | 4 ++--
>  1 file changed, 2 insertions(+), 2 deletions(-)
>
> diff --git a/accel/tcg/cputlb.c b/accel/tcg/cputlb.c
> index e72415a882..bcbcc4db6c 100644
> --- a/accel/tcg/cputlb.c
> +++ b/accel/tcg/cputlb.c
> @@ -765,7 +765,7 @@ static uint64_t io_readx(CPUArchState *env, CPUIOTLBEntry *iotlbentry,
>
>      cpu->mem_io_vaddr = addr;
>
> -    if (mr->global_locking) {
> +    if (mr->global_locking && !qemu_mutex_iothread_locked()) {
>          qemu_mutex_lock_iothread();
>          locked = true;
>      }
> @@ -800,7 +800,7 @@ static void io_writex(CPUArchState *env, CPUIOTLBEntry *iotlbentry,
>      cpu->mem_io_vaddr = addr;
>      cpu->mem_io_pc = retaddr;
>
> -    if (mr->global_locking) {
> +    if (mr->global_locking && !qemu_mutex_iothread_locked()) {
>          qemu_mutex_lock_iothread();
>          locked = true;
>      }
> --
> 2.14.1

Seems to fix the similar problem that I ran into with my work-in-progress
v8M blxns/secure-return code.

Reviewed-by: Peter Maydell <peter.maydell@linaro.org>

thanks
-- PMM

Re: [Qemu-devel] [PATCH] accel/tcg/cputlb: avoid recursive BQL (fixes #1706296)

[Richard Henderson]

On 09/21/2017 04:06 AM, Alex Bennée wrote:
> The mmio path (see exec.c:prepare_mmio_access) already protects itself
> against recursive locking and it makes sense to do the same for
> io_readx/writex. Otherwise any helper running in the BQL context will
> assert when it attempts to write to device memory as in the case of
> the bug report.
> 
> Signed-off-by: Alex Bennée <alex.bennee@linaro.org>
> CC: Richard Jones <rjones@redhat.com>
> CC: Paolo Bonzini <bonzini@gnu.org>
> ---
>  accel/tcg/cputlb.c | 4 ++--
>  1 file changed, 2 insertions(+), 2 deletions(-)

Reviewed-by: Richard Henderson <richard.henderson@linaro.org>

I guess I'll queue this through tcg-next?


r~

Re: [Qemu-devel] [PATCH] accel/tcg/cputlb: avoid recursive BQL (fixes #1706296)

[Michael Roth]

Quoting Alex Bennée (2017-09-21 06:06:25)
> The mmio path (see exec.c:prepare_mmio_access) already protects itself
> against recursive locking and it makes sense to do the same for
> io_readx/writex. Otherwise any helper running in the BQL context will
> assert when it attempts to write to device memory as in the case of
> the bug report.
> 
> Signed-off-by: Alex Bennée <alex.bennee@linaro.org>
> CC: Richard Jones <rjones@redhat.com>
> CC: Paolo Bonzini <bonzini@gnu.org>

FYI: this patch has been tagged for stable 2.10.1, but is not yet
upstream. Patch freeze for 2.10.1 is September 27th.

> ---
>  accel/tcg/cputlb.c | 4 ++--
>  1 file changed, 2 insertions(+), 2 deletions(-)
> 
> diff --git a/accel/tcg/cputlb.c b/accel/tcg/cputlb.c
> index e72415a882..bcbcc4db6c 100644
> --- a/accel/tcg/cputlb.c
> +++ b/accel/tcg/cputlb.c
> @@ -765,7 +765,7 @@ static uint64_t io_readx(CPUArchState *env, CPUIOTLBEntry *iotlbentry,
> 
>      cpu->mem_io_vaddr = addr;
> 
> -    if (mr->global_locking) {
> +    if (mr->global_locking && !qemu_mutex_iothread_locked()) {
>          qemu_mutex_lock_iothread();
>          locked = true;
>      }
> @@ -800,7 +800,7 @@ static void io_writex(CPUArchState *env, CPUIOTLBEntry *iotlbentry,
>      cpu->mem_io_vaddr = addr;
>      cpu->mem_io_pc = retaddr;
> 
> -    if (mr->global_locking) {
> +    if (mr->global_locking && !qemu_mutex_iothread_locked()) {
>          qemu_mutex_lock_iothread();
>          locked = true;
>      }
> -- 
> 2.14.1
> 
>