Re: [Qemu-devel] [PATCH v3 4/6] target/mips: hold BQL in mips_vpe_wake()

[Alex Bennée]

Aleksandar Markovic <address@hidden> writes:

> From: Goran Ferenc <address@hidden>
>
> Hold BQL whenever mips_vpe_wake() is invoked.
>
> Without this patch, MIPS MT with MTTCG enabled triggers an abort in
> tcg_handle_interrupt() due to an unlocked access to cpu_interrupt().
> This patch makes sure that the BQL is held in this case.

Just a reminder that this can go away in with Emilio's per-CPU locks patch.

>
> Signed-off-by: Goran Ferenc <address@hidden>
> Signed-off-by: Miodrag Dinic <address@hidden>
> Signed-off-by: Aleksandar Markovic <address@hidden>
> Acked-by: Alex Bennée <address@hidden>
> Reviewed-by: Alex Bennée <address@hidden>
> ---
>  target/mips/op_helper.c | 3 +++
>  1 file changed, 3 insertions(+)
>
> diff --git a/target/mips/op_helper.c b/target/mips/op_helper.c
> index 8c53b3b..0f272a5 100644
> --- a/target/mips/op_helper.c
> +++ b/target/mips/op_helper.c
> @@ -17,6 +17,7 @@
>   * License along with this library; if not, see 
> <http://www.gnu.org/licenses/>.
>   */
>  #include "qemu/osdep.h"
> +#include "qemu/main-loop.h"
>  #include "cpu.h"
>  #include "internal.h"
>  #include "qemu/host-utils.h"
> @@ -638,7 +639,9 @@ static inline void mips_vpe_wake(MIPSCPU *c)
>      /* Don't set ->halted = 0 directly, let it be done via cpu_has_work
>         because there might be other conditions that state that c should
>         be sleeping.  */
> +    qemu_mutex_lock_iothread();
>      cpu_interrupt(CPU(c), CPU_INTERRUPT_WAKE);
> +    qemu_mutex_unlock_iothread();
>  }
>
>  static inline void mips_vpe_sleep(MIPSCPU *cpu)


--
Alex Bennée