If a message is sent to multiple destinations, each destination might request a different set of metadata. Hence, we cannot rely on each one requesting at least a single bit.
Fix the message exporter to properly reset the metadata iovec on each emit-iteration. Otherwise, we might end up scanning random heap memory if a following destination does not request metadata. Reported-by: Jan Alexander Steffens <[email protected]> Signed-off-by: David Herrmann <[email protected]> --- Hi Greg One fall-out from the rewrite of the message importer. Reported by Jan. This fixes some issues with metadata on messages targeted at more than one destination. Reproducible by running a debug-monitor (which thus gets any messages, additionally to the normal receiver). Thanks David ipc/kdbus/message.c | 10 ++++++++++ 1 file changed, 10 insertions(+) diff --git a/ipc/kdbus/message.c b/ipc/kdbus/message.c index 64763ea..3520f45 100644 --- a/ipc/kdbus/message.c +++ b/ipc/kdbus/message.c @@ -946,6 +946,16 @@ struct kdbus_pool_slice *kdbus_staging_emit(struct kdbus_staging *staging, ++v; msg_size = KDBUS_ALIGN8(msg_size) + meta_size; + } else { + /* metadata items */ + v->iov_len = 0; + v->iov_base = (void __user *)zeros; + ++v; + + /* padding after metadata */ + v->iov_len = 0; + v->iov_base = (void __user *)zeros; + ++v; } /* ... payload iovecs are already filled in ... */ -- 2.4.5 -- To unsubscribe from this list: send the line "unsubscribe linux-kernel" in the body of a message to [email protected] More majordomo info at http://vger.kernel.org/majordomo-info.html Please read the FAQ at http://www.tux.org/lkml/
