From: Yann Droneaud <ydroneaud-RlY5vtjFyJ3QT0dZR+AlfA@public.gmane.org>
To: David Ahern <david.ahern-QHcLZuEGTsvQT0dZR+AlfA@public.gmane.org>
Cc: linux-rdma-u79uwXL29TY76Z2rM5mHXA@public.gmane.org,
jgunthorpe-ePGOBjL8dl3ta4EC/59zMFaTQe2KTcn/@public.gmane.org
Subject: Re: [PATCH v2] IB/core: Fix unaligned accesses
Date: Fri, 01 May 2015 10:45:25 +0200 [thread overview]
Message-ID: <1430469925.2957.14.camel@opteya.com> (raw)
In-Reply-To: <1430448168-38479-1-git-send-email-david.ahern-QHcLZuEGTsvQT0dZR+AlfA@public.gmane.org>
Hi,
Le jeudi 30 avril 2015 à 22:42 -0400, David Ahern a écrit :
> Addresses the following kernel logs seen during boot of sparc systems:
>
> Kernel unaligned access at TPC[103bce50] cm_find_listen+0x34/0xf8 [ib_cm]
> Kernel unaligned access at TPC[103bce50] cm_find_listen+0x34/0xf8 [ib_cm]
> Kernel unaligned access at TPC[103bce50] cm_find_listen+0x34/0xf8 [ib_cm]
> Kernel unaligned access at TPC[103bce50] cm_find_listen+0x34/0xf8 [ib_cm]
> Kernel unaligned access at TPC[103bce50] cm_find_listen+0x34/0xf8 [ib_cm]
>
> Changed signature of cm_mask_copy from u8 to u32 as suggested by Jason
> Gunthorpe.
> Signed-off-by: David Ahern <david.ahern-QHcLZuEGTsvQT0dZR+AlfA@public.gmane.org>
> ---
> v2
> - updated per Jason's comments
>
> drivers/infiniband/core/cm.c | 17 ++++++++---------
> include/rdma/ib_cm.h | 4 ++--
> 2 files changed, 10 insertions(+), 11 deletions(-)
>
> diff --git a/drivers/infiniband/core/cm.c b/drivers/infiniband/core/cm.c
> index e28a494e2a3a..cbbf741987b3 100644
> --- a/drivers/infiniband/core/cm.c
> +++ b/drivers/infiniband/core/cm.c
> @@ -437,20 +437,19 @@ static struct cm_id_private * cm_acquire_id(__be32 local_id, __be32 remote_id)
> return cm_id_priv;
> }
>
> -static void cm_mask_copy(u8 *dst, u8 *src, u8 *mask)
> +static void cm_mask_copy(u32 *dst, u32 *src, u32 *mask)
> {
> int i;
>
> - for (i = 0; i < IB_CM_COMPARE_SIZE / sizeof(unsigned long); i++)
> - ((unsigned long *) dst)[i] = ((unsigned long *) src)[i] &
> - ((unsigned long *) mask)[i];
> + for (i = 0; i < IB_CM_COMPARE_SIZE / sizeof(u32); i++)
> + dst[i] = src[i] & mask[i];
> }
>
> static int cm_compare_data(struct ib_cm_compare_data *src_data,
> struct ib_cm_compare_data *dst_data)
> {
> - u8 src[IB_CM_COMPARE_SIZE];
> - u8 dst[IB_CM_COMPARE_SIZE];
> + u32 src[IB_CM_COMPARE_SIZE];
> + u32 dst[IB_CM_COMPARE_SIZE];
>
You have to change IB_CM_COMPARE_SIZE, as it's going to allocate more
bytes than necessary.
Perhaps renaming IB_CM_COMPARE_SIZE to IB_CM_COMPARE_COUNT would be
good, then remove the division by sizeof(u32).
> if (!src_data || !dst_data)
> return 0;
> @@ -460,10 +459,10 @@ static int cm_compare_data(struct ib_cm_compare_data *src_data,
> return memcmp(src, dst, IB_CM_COMPARE_SIZE);
> }
>
> -static int cm_compare_private_data(u8 *private_data,
> +static int cm_compare_private_data(u32 *private_data,
> struct ib_cm_compare_data *dst_data)
> {
> - u8 src[IB_CM_COMPARE_SIZE];
> + u32 src[IB_CM_COMPARE_SIZE];
>
> if (!dst_data)
> return 0;
> @@ -546,7 +545,7 @@ static struct cm_id_private * cm_find_listen(struct ib_device *device,
>
> while (node) {
> cm_id_priv = rb_entry(node, struct cm_id_private, service_node);
> - data_cmp = cm_compare_private_data(private_data,
> + data_cmp = cm_compare_private_data((u32 *) private_data,
> cm_id_priv->compare_data);
> if ((cm_id_priv->id.service_mask & service_id) ==
> cm_id_priv->id.service_id &&
> diff --git a/include/rdma/ib_cm.h b/include/rdma/ib_cm.h
> index 0e3ff30647d5..0dd576f40c83 100644
> --- a/include/rdma/ib_cm.h
> +++ b/include/rdma/ib_cm.h
> @@ -337,8 +337,8 @@ void ib_destroy_cm_id(struct ib_cm_id *cm_id);
> #define IB_SDP_SERVICE_ID_MASK cpu_to_be64(0xFFFFFFFFFFFF0000ULL)
>
> struct ib_cm_compare_data {
> - u8 data[IB_CM_COMPARE_SIZE];
> - u8 mask[IB_CM_COMPARE_SIZE];
> + u32 data[IB_CM_COMPARE_SIZE];
> + u32 mask[IB_CM_COMPARE_SIZE];
> };
>
> /**
Anyway, you might want to compile this part of the kernel with
-Wcast-align and ignore the false positive (in particular
every other use of container_of()).
Regards.
--
Yann Droneaud
OPTEYA
--
To unsubscribe from this list: send the line "unsubscribe linux-rdma" in
the body of a message to majordomo-u79uwXL29TY76Z2rM5mHXA@public.gmane.org
More majordomo info at http://vger.kernel.org/majordomo-info.html
next prev parent reply other threads:[~2015-05-01 8:45 UTC|newest]
Thread overview: 8+ messages / expand[flat|nested] mbox.gz Atom feed top
2015-05-01 2:42 [PATCH v2] IB/core: Fix unaligned accesses David Ahern
[not found] ` <1430448168-38479-1-git-send-email-david.ahern-QHcLZuEGTsvQT0dZR+AlfA@public.gmane.org>
2015-05-01 8:45 ` Yann Droneaud [this message]
[not found] ` <1430469925.2957.14.camel-RlY5vtjFyJ3QT0dZR+AlfA@public.gmane.org>
2015-05-01 14:27 ` David Ahern
2015-05-01 9:50 ` Bart Van Assche
[not found] ` <55434C62.6050604-XdAiOPVOjttBDgjK7y7TUQ@public.gmane.org>
2015-05-01 14:28 ` David Ahern
2015-05-01 16:24 ` Jason Gunthorpe
[not found] ` <20150501162413.GC8531-ePGOBjL8dl3ta4EC/59zMFaTQe2KTcn/@public.gmane.org>
2015-05-01 16:33 ` David Ahern
[not found] ` <5543AACF.7060502-QHcLZuEGTsvQT0dZR+AlfA@public.gmane.org>
2015-05-01 16:57 ` Jason Gunthorpe
Reply instructions:
You may reply publicly to this message via plain-text email
using any one of the following methods:
* Save the following mbox file, import it into your mail client,
and reply-to-all from there: mbox
Avoid top-posting and favor interleaved quoting:
https://en.wikipedia.org/wiki/Posting_style#Interleaved_style
* Reply using the --to, --cc, and --in-reply-to
switches of git-send-email(1):
git send-email \
--in-reply-to=1430469925.2957.14.camel@opteya.com \
--to=ydroneaud-rly5vtjfyj3qt0dzr+alfa@public.gmane.org \
--cc=david.ahern-QHcLZuEGTsvQT0dZR+AlfA@public.gmane.org \
--cc=jgunthorpe-ePGOBjL8dl3ta4EC/59zMFaTQe2KTcn/@public.gmane.org \
--cc=linux-rdma-u79uwXL29TY76Z2rM5mHXA@public.gmane.org \
/path/to/YOUR_REPLY
https://kernel.org/pub/software/scm/git/docs/git-send-email.html
* If your mail client supports setting the In-Reply-To header
via mailto: links, try the mailto: link
Be sure your reply has a Subject: header at the top and a blank line
before the message body.
This is a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox