[Qemu-devel] [PATCH 1/2] dma-helpers: Initialize DMAAIOCB in_cancel flag

[Qemu-devel] [PATCH 1/2] dma-helpers: Initialize DMAAIOCB in_cancel flag

[Peter Maydell]

Initialize the dbs->in_cancel flag in dma_bdrv_io(), since qemu_aio_get()
does not return zero-initialized memory. Spotted by the clang sanitizer
(which complained when the value loaded in dma_complete() was not valid
for a bool type); this might have resulted in leaking the AIO block.

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

 dma-helpers.c | 1 +
 1 file changed, 1 insertion(+)

diff --git a/dma-helpers.c b/dma-helpers.c
index c9620a5..5f421e9 100644
--- a/dma-helpers.c
+++ b/dma-helpers.c
@@ -213,6 +213,7 @@ BlockDriverAIOCB *dma_bdrv_io(
     dbs->sg_cur_index = 0;
     dbs->sg_cur_byte = 0;
     dbs->dir = dir;
+    dbs->in_cancel = false;
     dbs->io_func = io_func;
     dbs->bh = NULL;
     qemu_iovec_init(&dbs->iov, sg->nsg);
-- 
1.9.0

Re: [Qemu-devel] [PATCH 1/2] dma-helpers: Initialize DMAAIOCB in_cancel flag

[Peter Maydell]

On 28 March 2014 14:22, Peter Maydell <peter.maydell@linaro.org> wrote:
> Initialize the dbs->in_cancel flag in dma_bdrv_io(), since qemu_aio_get()
> does not return zero-initialized memory. Spotted by the clang sanitizer
> (which complained when the value loaded in dma_complete() was not valid
> for a bool type); this might have resulted in leaking the AIO block.

Please ignore the 1/2 in the header -- this is a single
standalone patch, I just fluffed the git-format-patch command.

thanks
-- PMM

Re: [Qemu-devel] [PATCH 1/2] dma-helpers: Initialize DMAAIOCB in_cancel flag

[Peter Maydell]

On 28 March 2014 14:22, Peter Maydell <peter.maydell@linaro.org> wrote:
> Initialize the dbs->in_cancel flag in dma_bdrv_io(), since qemu_aio_get()
> does not return zero-initialized memory. Spotted by the clang sanitizer
> (which complained when the value loaded in dma_complete() was not valid
> for a bool type); this might have resulted in leaking the AIO block.
>
> Signed-off-by: Peter Maydell <peter.maydell@linaro.org>
> ---
>
>  dma-helpers.c | 1 +
>  1 file changed, 1 insertion(+)
>
> diff --git a/dma-helpers.c b/dma-helpers.c
> index c9620a5..5f421e9 100644
> --- a/dma-helpers.c
> +++ b/dma-helpers.c
> @@ -213,6 +213,7 @@ BlockDriverAIOCB *dma_bdrv_io(
>      dbs->sg_cur_index = 0;
>      dbs->sg_cur_byte = 0;
>      dbs->dir = dir;
> +    dbs->in_cancel = false;
>      dbs->io_func = io_func;
>      dbs->bh = NULL;
>      qemu_iovec_init(&dbs->iov, sg->nsg);
> --
> 1.9.0

Ping, or do we think this isn't worthwhile for 2.0?

(I guess I should have stuck a for-2.0 in the subject.)

thanks
-- PMM

Re: [Qemu-devel] [PATCH 1/2] dma-helpers: Initialize DMAAIOCB in_cancel flag

[Stefan Hajnoczi]

On Fri, Mar 28, 2014 at 02:22:49PM +0000, Peter Maydell wrote:
> Initialize the dbs->in_cancel flag in dma_bdrv_io(), since qemu_aio_get()
> does not return zero-initialized memory. Spotted by the clang sanitizer
> (which complained when the value loaded in dma_complete() was not valid
> for a bool type); this might have resulted in leaking the AIO block.
> 
> Signed-off-by: Peter Maydell <peter.maydell@linaro.org>
> ---
> 
>  dma-helpers.c | 1 +
>  1 file changed, 1 insertion(+)
> 
> diff --git a/dma-helpers.c b/dma-helpers.c
> index c9620a5..5f421e9 100644
> --- a/dma-helpers.c
> +++ b/dma-helpers.c
> @@ -213,6 +213,7 @@ BlockDriverAIOCB *dma_bdrv_io(
>      dbs->sg_cur_index = 0;
>      dbs->sg_cur_byte = 0;
>      dbs->dir = dir;
> +    dbs->in_cancel = false;
>      dbs->io_func = io_func;
>      dbs->bh = NULL;
>      qemu_iovec_init(&dbs->iov, sg->nsg);

Worth merging for 2.0.

Reviewed-by: Stefan Hajnoczi <stefanha@redhat.com>

Re: [Qemu-devel] [PATCH 1/2] dma-helpers: Initialize DMAAIOCB in_cancel flag

[Kevin Wolf]

Am 28.03.2014 um 15:22 hat Peter Maydell geschrieben:
> Initialize the dbs->in_cancel flag in dma_bdrv_io(), since qemu_aio_get()
> does not return zero-initialized memory. Spotted by the clang sanitizer
> (which complained when the value loaded in dma_complete() was not valid
> for a bool type); this might have resulted in leaking the AIO block.
> 
> Signed-off-by: Peter Maydell <peter.maydell@linaro.org>

Thanks, applied to the block branch.

Kevin