Message ID | 20230601161622.1808135-4-ckeepax@opensource.cirrus.com (mailing list archive) |
---|---|
State | Superseded |
Headers | show |
Series | [1/4] soundwire: stream: Add missing clear of alloc_slave_rt | expand |
On 6/1/23 11:16, Charles Keepax wrote: > Every error path in do_bank_switch prints an error message so there is no > need for the callers to also print error messages. Indeed in multi-master > cases these extra messages are confusing because they print out against a > random bus device whereas the do_bank_switch messages print against the bus > that actually failed. Errm, what? There is no way to know which bus failed in multi-master mode, and the 'stream' can span multiple buses so the use of pr_err was intentional. There's just no other way to report issues, and I don't see why one would remove such logs and fail silently. I just don't get what you are trying to address. > This also allows clean up of a couple of if's and variable initialisations > that were only there to silence potentially uninitialised variable > warnings on the bus variable. That should be a separate patch IMHO. > Signed-off-by: Charles Keepax <ckeepax@opensource.cirrus.com> > --- > drivers/soundwire/stream.c | 26 +++++--------------------- > 1 file changed, 5 insertions(+), 21 deletions(-) > > diff --git a/drivers/soundwire/stream.c b/drivers/soundwire/stream.c > index 248ab243ec6e4..b5c7a52aac19e 100644 > --- a/drivers/soundwire/stream.c > +++ b/drivers/soundwire/stream.c > @@ -1338,7 +1338,7 @@ static int _sdw_prepare_stream(struct sdw_stream_runtime *stream, > bool update_params) > { > struct sdw_master_runtime *m_rt; > - struct sdw_bus *bus = NULL; > + struct sdw_bus *bus; > struct sdw_master_prop *prop; > struct sdw_bus_params params; > int ret; > @@ -1380,16 +1380,9 @@ static int _sdw_prepare_stream(struct sdw_stream_runtime *stream, > } > } > > - if (!bus) { > - pr_err("Configuration error in %s\n", __func__); > - return -EINVAL; > - } > - > ret = do_bank_switch(stream); > - if (ret < 0) { > - pr_err("%s: do_bank_switch failed: %d\n", __func__, ret); > + if (ret < 0) > goto restore_params; > - } > > list_for_each_entry(m_rt, &stream->master_list, stream_node) { > bus = m_rt->bus; > @@ -1465,7 +1458,7 @@ EXPORT_SYMBOL(sdw_prepare_stream); > static int _sdw_enable_stream(struct sdw_stream_runtime *stream) > { > struct sdw_master_runtime *m_rt; > - struct sdw_bus *bus = NULL; > + struct sdw_bus *bus; > int ret; > > /* Enable Master(s) and Slave(s) port(s) associated with stream */ > @@ -1488,16 +1481,9 @@ static int _sdw_enable_stream(struct sdw_stream_runtime *stream) > } > } > > - if (!bus) { > - pr_err("Configuration error in %s\n", __func__); > - return -EINVAL; > - } > - > ret = do_bank_switch(stream); > - if (ret < 0) { > - pr_err("%s: do_bank_switch failed: %d\n", __func__, ret); > + if (ret < 0) > return ret; > - } > > stream->state = SDW_STREAM_ENABLED; > return 0; > @@ -1571,10 +1557,8 @@ static int _sdw_disable_stream(struct sdw_stream_runtime *stream) > } > > ret = do_bank_switch(stream); > - if (ret < 0) { > - pr_err("%s: do_bank_switch failed: %d\n", __func__, ret); > + if (ret < 0) > return ret; > - } > > /* make sure alternate bank (previous current) is also disabled */ > list_for_each_entry(m_rt, &stream->master_list, stream_node) {
On Thu, Jun 01, 2023 at 11:34:10AM -0500, Pierre-Louis Bossart wrote: > On 6/1/23 11:16, Charles Keepax wrote: > > Every error path in do_bank_switch prints an error message so there is no > > need for the callers to also print error messages. Indeed in multi-master > > cases these extra messages are confusing because they print out against a > > random bus device whereas the do_bank_switch messages print against the bus > > that actually failed. > > Errm, what? > > There is no way to know which bus failed in multi-master mode, and the > 'stream' can span multiple buses so the use of pr_err was intentional. Apologies this is the commit message not quite keeping pace with the code base. Originally when I wrote the patch the error message after do_bank_switch were a "dev_err(bus->dev", that was then fixed up in commit d014688eb373 ("soundwire: stream: remove bus->dev from logs on multiple buses"). > There's just no other way to report issues, and I don't see why one > would remove such logs and fail silently. > > I just don't get what you are trying to address. > The current code would say produce something like: Bank switch failed: -5 _sdw_prepare_stream: do_bank_switch failed: -5 I am sensing you are keen to keep both error messages, so fair enough I will drop that. Although worth noting originally before that patch I mention above this would have been: Bank switch failed: -5 do_bank_switch failed: -5 Which is really what I was attempting to address, that is clearly redundant. Now with the addition of the function in the print I guess it is slightly less redundant. > > This also allows clean up of a couple of if's and variable initialisations > > that were only there to silence potentially uninitialised variable > > warnings on the bus variable. > > That should be a separate patch IMHO. > I will trim the patch down to leave the duplicate error messages and just remove those bits. Thanks, Charles
diff --git a/drivers/soundwire/stream.c b/drivers/soundwire/stream.c index 248ab243ec6e4..b5c7a52aac19e 100644 --- a/drivers/soundwire/stream.c +++ b/drivers/soundwire/stream.c @@ -1338,7 +1338,7 @@ static int _sdw_prepare_stream(struct sdw_stream_runtime *stream, bool update_params) { struct sdw_master_runtime *m_rt; - struct sdw_bus *bus = NULL; + struct sdw_bus *bus; struct sdw_master_prop *prop; struct sdw_bus_params params; int ret; @@ -1380,16 +1380,9 @@ static int _sdw_prepare_stream(struct sdw_stream_runtime *stream, } } - if (!bus) { - pr_err("Configuration error in %s\n", __func__); - return -EINVAL; - } - ret = do_bank_switch(stream); - if (ret < 0) { - pr_err("%s: do_bank_switch failed: %d\n", __func__, ret); + if (ret < 0) goto restore_params; - } list_for_each_entry(m_rt, &stream->master_list, stream_node) { bus = m_rt->bus; @@ -1465,7 +1458,7 @@ EXPORT_SYMBOL(sdw_prepare_stream); static int _sdw_enable_stream(struct sdw_stream_runtime *stream) { struct sdw_master_runtime *m_rt; - struct sdw_bus *bus = NULL; + struct sdw_bus *bus; int ret; /* Enable Master(s) and Slave(s) port(s) associated with stream */ @@ -1488,16 +1481,9 @@ static int _sdw_enable_stream(struct sdw_stream_runtime *stream) } } - if (!bus) { - pr_err("Configuration error in %s\n", __func__); - return -EINVAL; - } - ret = do_bank_switch(stream); - if (ret < 0) { - pr_err("%s: do_bank_switch failed: %d\n", __func__, ret); + if (ret < 0) return ret; - } stream->state = SDW_STREAM_ENABLED; return 0; @@ -1571,10 +1557,8 @@ static int _sdw_disable_stream(struct sdw_stream_runtime *stream) } ret = do_bank_switch(stream); - if (ret < 0) { - pr_err("%s: do_bank_switch failed: %d\n", __func__, ret); + if (ret < 0) return ret; - } /* make sure alternate bank (previous current) is also disabled */ list_for_each_entry(m_rt, &stream->master_list, stream_node) {
Every error path in do_bank_switch prints an error message so there is no need for the callers to also print error messages. Indeed in multi-master cases these extra messages are confusing because they print out against a random bus device whereas the do_bank_switch messages print against the bus that actually failed. This also allows clean up of a couple of if's and variable initialisations that were only there to silence potentially uninitialised variable warnings on the bus variable. Signed-off-by: Charles Keepax <ckeepax@opensource.cirrus.com> --- drivers/soundwire/stream.c | 26 +++++--------------------- 1 file changed, 5 insertions(+), 21 deletions(-)