On 09/01/2019 01:51, Kuninori Morimoto wrote: > > Hi Jon > > Thank you for your help > >>> Yes so this does workaround the problem. However, per my previous >>> comments, I would like to explore whether it is necessary to allocate >>> the platform link component or if it can be static. > > OK, thanks. > It *will* be static, but not yet. > Thank you for your patch. > I guess you worry about allocated memory leak when failed case ? > It is using devm_kzalloc(), so all allocated memory will be automatically > freed when card was detached. > But indeed if driver gots -EPROBE_DEFER many times, > it will allocate many unused platform. No. Offline you had suggested using kmalloc and not devm_kzalloc and so I was worried in that case about a memory leak. Right now I am only concerned about an invalid pointer that is not being handled correctly. > Here is the background of snd_soc_init_platform. > > Legacy dai_link was xxx_name/xxx_of_node style, > but multi codec support starts to use snd_soc_dai_link_component style. > OTOH Lars-Petter is thinking that current ALSA SoC is not good match > for modern sound device. > I guess, we need "multi xxx" support as 1st step for modern sound device. > "multi codec" is already supported, > "multi cpu" patch was posted, but not yet accepted (or rejected ??). > "multi platform" is no plan (?). I would like someone to explain what multi-platform means? Even if a soundcard supports multiple platforms, there is only one platform you are using at any time so ... > These want to use snd_soc_dai_link_component style, > because it is nice for multi xxx support style, I think. > I think no one is planing for "multi platform" so far, thus, > I posted snd_soc_dai_link_component style for it. > # Maybe it should have num_platform, too. > # all driver will have .num_platform = 1, thus I didn't added. > # maybe it was my fault... ... I don't understand why you would ever need a 'num_platform' as the machine driver just needs to understand which platform is using it at any given time. Right? > The reason why platform/codec is allocating/copying by snd_soc_init_xxx > so far is that it is glue for > xxx_name/xxx_of_node (legacy style) <-> snd_soc_init_platform (modern style). > > I want to which to modern style immediately and remove legacy style. > But as you know, we have too many ALSA SoC drivers now. > So, if I posted "switch legacy style to modern style" patch > for each (= for codec, for platform, for cpu), it will be patch-bomb, > and Lars-Petter/Mark don't like it. > Thus, I'm waiting "multi CPU" support patch. Sorry, I still don't understand the dependency on the multi CPU and why we need to wait. > If CPU/Codec/Platform can be snd_soc_init_platform style, > then, we can switch to modern style for all drivers. > Then, all driver will have *static* platform. > > # So, I guess if your driver can switch to use > # snd_soc_init_platform style directly, your problem can gone ? Yes that is an alternative and I can convert all the Tegra machine drivers to use this now. However, that will not solve the problem for non-Tegra devices and everyone will have to do this. Cheers Jon -- nvpublic
