[pulseaudio-tickets] [Bug 88452] pulseaudio "killed" my qemu instance (assertion hit)

[bugzilla-daemon at freedesktop.org]

https://bugs.freedesktop.org/show_bug.cgi?id=88452

--- Comment #16 from David Henningsson <david.henningsson at canonical.com> ---
I think the section below /* Frame payload available */ in do_read is broken w
r t what happens if the packet is split up.

Look how this offset parameter does not take re->index into account:

offset = (int64_t) ((((uint64_t)
ntohl(re->descriptor[PA_PSTREAM_DESCRIPTOR_OFFSET_HI])) << 32) | (((uint64_t)
ntohl(re->descriptor[PA_PSTREAM_DESCRIPTOR_OFFSET_LO]))));

Now, that happens to work if and only if
 1) the seek mode is PA_SEEK_RELATIVE and
 2) the previously packet was successfully pushed into the queue.

If any of these two fails, any memchunk except the first in a split will be
pushed into the wrong place in the queue.

Maybe there is some advantage in getting some of the data as quick as possible,
but I think this split will be too hairy to sort out correctly. My suggestion
is that we do not call receive_memblock_callback until the entire block is
received. Opinions?

-- 
You are receiving this mail because:
You are the QA Contact for the bug.
You are the assignee for the bug.
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.freedesktop.org/archives/pulseaudio-bugs/attachments/20150213/9037d5c3/attachment.html>