Asrc module convert #8416
Asrc module convert #8416
Conversation
andrula-song
force-pushed
the
asrc-module-convert
branch
from
4223d49 to
f0ea479
Compare
src/audio/asrc/asrc_ipc3.c
Outdated
| @@ -0,0 +1,1027 @@ | |||
| // SPDX-License-Identifier: BSD-3-Clause | |||
There was a problem hiding this comment.
can we extract common code?
There was a problem hiding this comment.
Is there any else to extract? Thanks for pointing out in detail.
There was a problem hiding this comment.
@lyakh , we can use another PR to do the extraction and ipc3/ipc4 isolation, this PR can only focus on module convertion.
andrula-song
force-pushed
the
asrc-module-convert
branch
3 times, most recently
from
40c21c2 to
128f2b3
Compare
andrula-song
force-pushed
the
asrc-module-convert
branch
2 times, most recently
from
9151025 to
cceaaa4
Compare
Add IPC4 DAI timestamp ops for ASRC. Signed-off-by: Andrula Song <[email protected]>
Add trigger ops in module_interface. Signed-off-by: Andrula Song <[email protected]>
Convert ASRC to module API and support ASRC IPC3 test with test bench. Signed-off-by: Andrula Song <[email protected]>
andrula-song
force-pushed
the
asrc-module-convert
branch
from
cceaaa4 to
35681ba
Compare
There was a problem hiding this comment.
@andrula-song ， please change status to open, this PR should be done now.
andrula-song
requested review from
ranj063,
jxstelter,
lgirdwood,
plbossart,
mmaka1,
lbetlej,
dbaluta and
kv2019i
as code owners
|
SOFCI TEST |
| src_obj->ring_buffers32[ch] = NULL; | ||
| rfree(buf_32); |
There was a problem hiding this comment.
Interesting, was this a memory violation fix? Please explain and move to a separate commit.
There was a problem hiding this comment.
ah, it is because the comment from @lyakh #8416 (comment)
There was a problem hiding this comment.
@singalsu the original order keeps a theoretical race window open: while you still have a pointer to a freed memory in a globally accessible persistent object, someone might access it at that time, causing a use-after-free bug scenario. Although we don't know any such specific possibilities here, but since we already have a temporary variable with that pointer, we could as well use it to protect against that race possibility. I.e.
free(obj->mem);
// if someone now accesses obj->mem, it's a use-after-free bug
obj->mem = NULL;
that one has a race window, whereas
x = obj->mem;
obj->mem = NULL;
free(x);
this one doesn't
There was a problem hiding this comment.
Thanks for elaborating @lyakh !
convert ASRC to module interface with IPC4.