[alsa-devel] [PATCH alsa-lib 1/4] ucm: Fix opening of master-configs by the card's longname

[Hans de Goede]

Hi,

On 19-11-2019 13:23, Jaroslav Kysela wrote:
> Dne 19. 11. 19 v 11:48 Hans de Goede napsal(a):
>> Prior to commit aba2260ae7b5 ("ucm: switch to ucm2 directory and v2 format,
>> keep backward compatibility").
>>
>> The filename build in parse_verb_file() was build like this:
>> <prefix>/<uc_mgr->conf_file_name>/<file>
>>
>> Where uc_mgr->conf_file_name would contain either the card_name or the
>> card's longname depending on the detection of a longname based config in
>> uc_mgr_import_master_config().
>>
>> While the filename used in load_master_config() was build like this:
>> <prefix>/<card_name>/<card_name>.conf
>>
>> And uc_mgr_import_master_config() first calls load_master_config()
>> with the card's longname and if that succeeds it overwrites
>> uc_mgr->conf_file_name with the longname so that the subsequent uses
>> of uc_mgr->conf_file_name in parse_verb_file() correctly use the longname.
>>
>> But the new configuration_filename() helper added in commit aba2260ae7b5
>> _always_ builds the filename like this:
>> <prefix>/<uc_mgr->conf_file_name>/<file><suffix>
>>
>> This breaks the loading of the master-config by its longname, as when
>> the longname is tried uc_mgr->conf_file_name still contains the card_name.
> 
> Hi Hans,
> 
> This new behaviour was intended for ucm2. I applied all four your patches, and added 10a63e093c4a98acfa9bcdfdd06938bcd874b008 on top which should keep the
> new behaviour for v2 configs.

Ah I see.

There is a bug in the commit you added on top though,
when load_master_config() gets called with longname == 1
and ALSA_CONFIG_UCM2_VAR is not set and ALSA_CONFIG_UCM_VAR
is set, then the if condition in the block is false, so it never executes:

+               if (getenv(ALSA_CONFIG_UCM2_VAR) || !getenv(ALSA_CONFIG_UCM_VAR)) {
+                       uc_mgr->conf_format = 2;
+                       configuration_filename(uc_mgr, filename, sizeof(filename),
+                                              uc_mgr->conf_file_name, card_name, ".conf");
+               }

Causing the next block to also not execute:

+               if (uc_mgr->conf_format >= 2 && access(filename, R_OK) != 0) {
+                       /* try the old ucm directory */
+                       uc_mgr->conf_format = 1;
+                       configuration_filename(uc_mgr, filename, sizeof(filename),
+                                              card_name, card_name, ".conf");
+                       if (access(filename, R_OK) != 0)
+                               return -ENOENT;
+               }

Causing load_master_config() to continue with an uninitialized filename.

I think you wanted the following condition for the second block:

                 if (uc_mgr->conf_format == 0 || access(filename, R_OK) != 0) {

Also don't you want the same behavor wrt forcing the conf_format based
on env. when trying the shortname ?

I think you can move the entire block under if (longname) to be unconditional,
since conf_file_name == card_name for the shortname case using the longname
path in the shortname case too is fine. Then you always call configuration_filename
with conf_format set and all the handling for conf_format not being set (and
for overriding it) can be dropped from configuration_filename() then.

> It's time to create the 1.2.1.1 bugfix alsa-lib release, I guess.

Yes that is a good idea (once the above is resolved).

Regards,

Hans