| Message ID | 1473081421-16555-18-git-send-email-peter.griffin@linaro.org (mailing list archive) |
|---|---|
| State | New, archived |
| Headers | show |
On 5 September 2016 at 14:16, Peter Griffin <peter.griffin@linaro.org> wrote: > ST_SLIM_REMOTEPROC must select REMOTEPROC, which exposes the following > recursive dependency. > From a humble skim through remoteproc, drm and a few other subsystems I think the above is wrong. All the drivers (outside of remoteproc), that I've seen, depend on the core component, they don't select it. Furthermore most places explicitly hide the drivers from the menu if the core component isn't enabled. Is there something that requires such a different/unusual behaviour in remoteproc ? Regards, Emil
Hi Emil, On Tue, 20 Sep 2016, Emil Velikov wrote: > On 5 September 2016 at 14:16, Peter Griffin <peter.griffin@linaro.org> wrote: > > ST_SLIM_REMOTEPROC must select REMOTEPROC, which exposes the following > > recursive dependency. > > > From a humble skim through remoteproc, drm and a few other subsystems > I think the above is wrong. All the drivers (outside of remoteproc), > that I've seen, depend on the core component, they don't select it. I will let Bjorn comment on the remoteproc subsystem Kconfig design, and why it is like it is. For this particular SLIM_RPROC I have added it to Kconfig in keeping with all the other drivers in the remoteproc subsystem which has exposed this recursive dependency issue. For this particular kconfig symbol a quick grep reveals more drivers in the kernel using 'select', than 'depend on' git grep "select VIRTIO" | wc -l 14 git grep "depends on VIRTIO" | wc -l 10 > Furthermore most places explicitly hide the drivers from the menu if > the core component isn't enabled. Remoteproc subsystem takes a different approach, the core code is only enabled if a driver which relies on it is enabled. This IMHO makes sense, as remoteproc is not widely used (only a few particular ARM SoC's). It is true that for subsystems which rely on the core component being explicitly enabled, they often tend to hide drivers which depend on it from the menu unless it is. This also makes sense. > > Is there something that requires such a different/unusual behaviour in > remoteproc ? > I'm not sure it is that unusual...looking at config USB, it selects USB_COMMON in mfd subsystem, client drivers select MFD_CORE. I've added Arnd to this thread, as I've seen lots of Kconfig patches from him recently, maybe he has some thoughts on whether this is the correct fix or not? regards, Peter.
On Tue, 20 Sep 2016, Peter Griffin <peter.griffin@linaro.org> wrote: > Hi Emil, > > On Tue, 20 Sep 2016, Emil Velikov wrote: > >> On 5 September 2016 at 14:16, Peter Griffin <peter.griffin@linaro.org> wrote: >> > ST_SLIM_REMOTEPROC must select REMOTEPROC, which exposes the following >> > recursive dependency. > > >> > >> From a humble skim through remoteproc, drm and a few other subsystems >> I think the above is wrong. All the drivers (outside of remoteproc), >> that I've seen, depend on the core component, they don't select it. > > I will let Bjorn comment on the remoteproc subsystem Kconfig design, and > why it is like it is. > > For this particular SLIM_RPROC I have added it to Kconfig in keeping with all > the other drivers in the remoteproc subsystem which has exposed this recursive > dependency issue. > > For this particular kconfig symbol a quick grep reveals more drivers in > the kernel using 'select', than 'depend on' > > git grep "select VIRTIO" | wc -l > 14 > > git grep "depends on VIRTIO" | wc -l > 10 > > >> Furthermore most places explicitly hide the drivers from the menu if >> the core component isn't enabled. > > Remoteproc subsystem takes a different approach, the core code is only enabled > if a driver which relies on it is enabled. This IMHO makes sense, as > remoteproc is not widely used (only a few particular ARM SoC's). > > It is true that for subsystems which rely on the core component being > explicitly enabled, they often tend to hide drivers which depend on it > from the menu unless it is. This also makes sense. > >> >> Is there something that requires such a different/unusual behaviour in >> remoteproc ? >> > > I'm not sure it is that unusual...looking at config USB, it selects USB_COMMON in > mfd subsystem, client drivers select MFD_CORE. > > I've added Arnd to this thread, as I've seen lots of Kconfig patches from him > recently, maybe he has some thoughts on whether this is the correct fix or not? Documentation/kbuild/kconfig-language.txt: Note: select should be used with care. select will force a symbol to a value without visiting the dependencies. By abusing select you are able to select a symbol FOO even if FOO depends on BAR that is not set. In general use select only for non-visible symbols (no prompts anywhere) and for symbols with no dependencies. That will limit the usefulness but on the other hand avoid the illegal configurations all over. People tend to abuse select because it's "convenient". If you depend, but some of your dependencies aren't met, you're in for some digging through Kconfig to find the missing deps. Just to make the option you want visible in menuconfig. If you instead select something with dependencies, it'll be right most of the time, and it's "convenient", until it breaks. (And hey, it usually breaks for someone else with some other config, so it's still convenient for you.) Perhaps kconfig should complain about selecting visible symbols and symbols with dependencies. BR, Jani.
On 20 September 2016 at 09:32, Peter Griffin <peter.griffin@linaro.org> wrote: > Hi Emil, > > On Tue, 20 Sep 2016, Emil Velikov wrote: > >> On 5 September 2016 at 14:16, Peter Griffin <peter.griffin@linaro.org> wrote: >> > ST_SLIM_REMOTEPROC must select REMOTEPROC, which exposes the following >> > recursive dependency. > > >> > >> From a humble skim through remoteproc, drm and a few other subsystems >> I think the above is wrong. All the drivers (outside of remoteproc), >> that I've seen, depend on the core component, they don't select it. > > I will let Bjorn comment on the remoteproc subsystem Kconfig design, and > why it is like it is. > > For this particular SLIM_RPROC I have added it to Kconfig in keeping with all > the other drivers in the remoteproc subsystem which has exposed this recursive > dependency issue. > > For this particular kconfig symbol a quick grep reveals more drivers in > the kernel using 'select', than 'depend on' > > git grep "select VIRTIO" | wc -l > 14 > > git grep "depends on VIRTIO" | wc -l > 10 > Might be worth taking a closer look into these at some point. > >> Furthermore most places explicitly hide the drivers from the menu if >> the core component isn't enabled. > > Remoteproc subsystem takes a different approach, the core code is only enabled > if a driver which relies on it is enabled. This IMHO makes sense, as > remoteproc is not widely used (only a few particular ARM SoC's). > > It is true that for subsystems which rely on the core component being > explicitly enabled, they often tend to hide drivers which depend on it > from the menu unless it is. This also makes sense. > >> >> Is there something that requires such a different/unusual behaviour in >> remoteproc ? >> > > I'm not sure it is that unusual...looking at config USB, it selects USB_COMMON in > mfd subsystem, client drivers select MFD_CORE. > On the USB case I'm not sure what the reasoning behind the USB vs USB_COMMON split. In seems that one could just fold them, but that's another topic. On the MFD side... it follows the select {,mis,ab}use. With one (the only one?) MFD driver not using/selecting MFD_CORE doing it's own version of mfd_add_devices... which could be reworked, possibly. > I've added Arnd to this thread, as I've seen lots of Kconfig patches from him > recently, maybe he has some thoughts on whether this is the correct fix or not? > Ack. Fwiw, I believe that the reasoning put by Jani is perfeclty reasonable, but it'll be great to hear others as well. Thanks Emil
On Wed 21 Sep 05:09 PDT 2016, Emil Velikov wrote: > On 20 September 2016 at 09:32, Peter Griffin <peter.griffin@linaro.org> wrote: > > Hi Emil, > > > > On Tue, 20 Sep 2016, Emil Velikov wrote: > > > >> On 5 September 2016 at 14:16, Peter Griffin <peter.griffin@linaro.org> wrote: > >> > ST_SLIM_REMOTEPROC must select REMOTEPROC, which exposes the following > >> > recursive dependency. > > > > > >> > > >> From a humble skim through remoteproc, drm and a few other subsystems > >> I think the above is wrong. All the drivers (outside of remoteproc), > >> that I've seen, depend on the core component, they don't select it. > > > > I will let Bjorn comment on the remoteproc subsystem Kconfig design, and > > why it is like it is. > > > > For this particular SLIM_RPROC I have added it to Kconfig in keeping with all > > the other drivers in the remoteproc subsystem which has exposed this recursive > > dependency issue. > > > > For this particular kconfig symbol a quick grep reveals more drivers in > > the kernel using 'select', than 'depend on' > > > > git grep "select VIRTIO" | wc -l > > 14 > > > > git grep "depends on VIRTIO" | wc -l > > 10 > > > Might be worth taking a closer look into these at some point. > The general idea here is that VIRTIO provides the "framework" and as such drivers implementing VIRTIO do select and drivers using virtio use depends. This is found in several places around the kernel. > > > >> Furthermore most places explicitly hide the drivers from the menu if > >> the core component isn't enabled. > > > > Remoteproc subsystem takes a different approach, the core code is only enabled > > if a driver which relies on it is enabled. This IMHO makes sense, as > > remoteproc is not widely used (only a few particular ARM SoC's). > > > > It is true that for subsystems which rely on the core component being > > explicitly enabled, they often tend to hide drivers which depend on it > > from the menu unless it is. This also makes sense. > > > >> > >> Is there something that requires such a different/unusual behaviour in > >> remoteproc ? > >> There's nothing unusual in remoteproc that forces us to stay with this model; however the parts related to the REMOTEPROC config is useless by themselves. Regards, Bjorn
Hi Jani, Sorry for the delay, I've been travelling last week. On Tue, 20 Sep 2016, Jani Nikula wrote: > On Tue, 20 Sep 2016, Peter Griffin <peter.griffin@linaro.org> wrote: > > Hi Emil, > > > > On Tue, 20 Sep 2016, Emil Velikov wrote: > > > >> On 5 September 2016 at 14:16, Peter Griffin <peter.griffin@linaro.org> wrote: > >> > ST_SLIM_REMOTEPROC must select REMOTEPROC, which exposes the following > >> > recursive dependency. > > > > > >> > > >> From a humble skim through remoteproc, drm and a few other subsystems > >> I think the above is wrong. All the drivers (outside of remoteproc), > >> that I've seen, depend on the core component, they don't select it. > > > > I will let Bjorn comment on the remoteproc subsystem Kconfig design, and > > why it is like it is. > > > > For this particular SLIM_RPROC I have added it to Kconfig in keeping with all > > the other drivers in the remoteproc subsystem which has exposed this recursive > > dependency issue. > > > > For this particular kconfig symbol a quick grep reveals more drivers in > > the kernel using 'select', than 'depend on' > > > > git grep "select VIRTIO" | wc -l > > 14 > > > > git grep "depends on VIRTIO" | wc -l > > 10 > > > > > >> Furthermore most places explicitly hide the drivers from the menu if > >> the core component isn't enabled. > > > > Remoteproc subsystem takes a different approach, the core code is only enabled > > if a driver which relies on it is enabled. This IMHO makes sense, as > > remoteproc is not widely used (only a few particular ARM SoC's). > > > > It is true that for subsystems which rely on the core component being > > explicitly enabled, they often tend to hide drivers which depend on it > > from the menu unless it is. This also makes sense. > > > >> > >> Is there something that requires such a different/unusual behaviour in > >> remoteproc ? > >> > > > > I'm not sure it is that unusual...looking at config USB, it selects USB_COMMON in > > mfd subsystem, client drivers select MFD_CORE. > > > > I've added Arnd to this thread, as I've seen lots of Kconfig patches from him > > recently, maybe he has some thoughts on whether this is the correct fix or not? > > > Documentation/kbuild/kconfig-language.txt: > > Note: > select should be used with care. select will force > a symbol to a value without visiting the dependencies. > By abusing select you are able to select a symbol FOO even > if FOO depends on BAR that is not set. > In general use select only for non-visible symbols > (no prompts anywhere) and for symbols with no dependencies. > That will limit the usefulness but on the other hand avoid > the illegal configurations all over. Thanks for the documentation link. I believe this proves this patch is an appropriate fix as VIRTIO is a non-visible symbol, and has no dependencies. In fact the help text for VIRTIO even states this option should be selected by any driver which implements virtio. > > People tend to abuse select because it's "convenient". If you depend, > but some of your dependencies aren't met, you're in for some digging > through Kconfig to find the missing deps. Just to make the option you > want visible in menuconfig. If you instead select something with > dependencies, it'll be right most of the time, and it's "convenient", > until it breaks. (And hey, it usually breaks for someone else with some > other config, so it's still convenient for you.) I'm sure they do but in this case it is actually the use of 'depends on' which has caused the breakage and inconvenience for somebody else and sadly this inconvienice is still on-going due to this patch not being applied or getting an acked-by from the appropriate maintainers. > > Perhaps kconfig should complain about selecting visible symbols and > symbols with dependencies. That sounds like it would be a useful addition. Is it possible to get this patch applied or an acked-by to avoid further delay to the fdma series? regards, Peter.
Hi Bjorn, On 27 September 2016 at 18:01, Bjorn Andersson <bjorn.andersson@linaro.org> wrote: > On Wed 21 Sep 05:09 PDT 2016, Emil Velikov wrote: > >> On 20 September 2016 at 09:32, Peter Griffin <peter.griffin@linaro.org> wrote: >> > Hi Emil, >> > >> > On Tue, 20 Sep 2016, Emil Velikov wrote: >> > >> >> On 5 September 2016 at 14:16, Peter Griffin <peter.griffin@linaro.org> wrote: >> >> > ST_SLIM_REMOTEPROC must select REMOTEPROC, which exposes the following >> >> > recursive dependency. >> > >> > >> >> > >> >> From a humble skim through remoteproc, drm and a few other subsystems >> >> I think the above is wrong. All the drivers (outside of remoteproc), >> >> that I've seen, depend on the core component, they don't select it. >> > >> > I will let Bjorn comment on the remoteproc subsystem Kconfig design, and >> > why it is like it is. >> > >> > For this particular SLIM_RPROC I have added it to Kconfig in keeping with all >> > the other drivers in the remoteproc subsystem which has exposed this recursive >> > dependency issue. >> > >> > For this particular kconfig symbol a quick grep reveals more drivers in >> > the kernel using 'select', than 'depend on' >> > >> > git grep "select VIRTIO" | wc -l >> > 14 >> > >> > git grep "depends on VIRTIO" | wc -l >> > 10 >> > >> Might be worth taking a closer look into these at some point. >> > > The general idea here is that VIRTIO provides the "framework" and as > such drivers implementing VIRTIO do select and drivers using virtio use > depends. > > This is found in several places around the kernel. > >> > >> >> Furthermore most places explicitly hide the drivers from the menu if >> >> the core component isn't enabled. >> > >> > Remoteproc subsystem takes a different approach, the core code is only enabled >> > if a driver which relies on it is enabled. This IMHO makes sense, as >> > remoteproc is not widely used (only a few particular ARM SoC's). >> > >> > It is true that for subsystems which rely on the core component being >> > explicitly enabled, they often tend to hide drivers which depend on it >> > from the menu unless it is. This also makes sense. >> > >> >> >> >> Is there something that requires such a different/unusual behaviour in >> >> remoteproc ? >> >> > > There's nothing unusual in remoteproc that forces us to stay with this > model; however the parts related to the REMOTEPROC config is useless by > themselves. > I'm afraid that the "supporting" arguments you're using are generic and not specific to remoteproc. Although as Jani mentioned and pointed to the documentation, remoteproc is {mis,ab}using 'select' leading to issues elsewhere. In the long term we might want to switch to 'select' and attribute kconfig like Jani suggested. But in the short term we want to avoid select-ing things just for our "convenience". Thanks Emil
On 6 October 2016 at 10:37, Peter Griffin <peter.griffin@linaro.org> wrote: > In fact the help text for VIRTIO even states this option should be selected > by any driver which implements virtio. > Almost but not quite. It says: "This option is selected by any driver which implements the virtio _bus_" REMOTEPROC obviously does that while the ST SLIM driver does not. Thus the latter should _not_ select, be that explicitly or implicitly via REMOTEPROC, the symbol. >> >> People tend to abuse select because it's "convenient". If you depend, >> but some of your dependencies aren't met, you're in for some digging >> through Kconfig to find the missing deps. Just to make the option you >> want visible in menuconfig. If you instead select something with >> dependencies, it'll be right most of the time, and it's "convenient", >> until it breaks. (And hey, it usually breaks for someone else with some >> other config, so it's still convenient for you.) > > I'm sure they do but in this case it is actually the use of 'depends on' > which has caused the breakage and inconvenience for somebody else and sadly this > inconvienice is still on-going due to this patch not being applied or getting an > acked-by from the appropriate maintainers. > Surely you're not saying that pre-existing driver following the documentation, is 'causing breakage' for a new driver {ab,mis}using a feature ? This reminds me an old saying: "If the shoe doesn’t fit, it doesn’t mean there is anything wrong with your feet." You seem to be suggesting the opposite ? >> >> Perhaps kconfig should complain about selecting visible symbols and >> symbols with dependencies. > > That sounds like it would be a useful addition. > > Is it possible to get this patch applied or an acked-by to avoid further delay > to the fdma series? > Please don't apply duct tape, especially where it's _not_ needed. $ sed -i s/select REMOTEPROC/depends on REMOTEPROC/ drivers/remoteproc/Kconfig ... will resolve things in the right place. The alternative will lead to random issues in other subsystems. Regards, Emil
Hi Emil, On Wed, 21 Sep 2016, Emil Velikov wrote: > On 20 September 2016 at 09:32, Peter Griffin <peter.griffin@linaro.org> wrote: > > Hi Emil, > > > > On Tue, 20 Sep 2016, Emil Velikov wrote: > > > >> On 5 September 2016 at 14:16, Peter Griffin <peter.griffin@linaro.org> wrote: > >> > ST_SLIM_REMOTEPROC must select REMOTEPROC, which exposes the following > >> > recursive dependency. > > > > > >> > > >> From a humble skim through remoteproc, drm and a few other subsystems > >> I think the above is wrong. All the drivers (outside of remoteproc), > >> that I've seen, depend on the core component, they don't select it. > > > > I will let Bjorn comment on the remoteproc subsystem Kconfig design, and > > why it is like it is. > > > > For this particular SLIM_RPROC I have added it to Kconfig in keeping with all > > the other drivers in the remoteproc subsystem which has exposed this recursive > > dependency issue. > > > > For this particular kconfig symbol a quick grep reveals more drivers in > > the kernel using 'select', than 'depend on' > > > > git grep "select VIRTIO" | wc -l > > 14 > > > > git grep "depends on VIRTIO" | wc -l > > 10 > > > Might be worth taking a closer look into these at some point. VIRTIO has no dependencies, and is a non visible symbol. Its Kconfig help also mentions that drivers should select it. > > > > >> Furthermore most places explicitly hide the drivers from the menu if > >> the core component isn't enabled. > > > > Remoteproc subsystem takes a different approach, the core code is only enabled > > if a driver which relies on it is enabled. This IMHO makes sense, as > > remoteproc is not widely used (only a few particular ARM SoC's). > > > > It is true that for subsystems which rely on the core component being > > explicitly enabled, they often tend to hide drivers which depend on it > > from the menu unless it is. This also makes sense. > > > >> > >> Is there something that requires such a different/unusual behaviour in > >> remoteproc ? > >> > > > > I'm not sure it is that unusual...looking at config USB, it selects USB_COMMON in > > mfd subsystem, client drivers select MFD_CORE. > > > On the USB case I'm not sure what the reasoning behind the USB vs > USB_COMMON split. In seems that one could just fold them, but that's > another topic. On the MFD side... it follows the select {,mis,ab}use. > With one (the only one?) MFD driver not using/selecting MFD_CORE doing > it's own version of mfd_add_devices... which could be reworked, > possibly. Much like selecting VIRTIO in this patch, MFD_CORE is a non visible symbol with no dependencies so it matches the documentation Jani referenced. I personally think the approach taken makes sense, as why would you want to have a CONFIG_MFD_CORE=y symbol being enabled unless you actually have a MFD device which uses it also enabled in your kernel? It seems to me a good solution to make the client drivers select the core component so that it only gets enabled if at least one driver requires it. This avoids having useless core code which will never be used compiled into the kernel and in the end a smaller kernel size for a given kernel configuration (better cache use etc etc). > > I've added Arnd to this thread, as I've seen lots of Kconfig patches from him > > recently, maybe he has some thoughts on whether this is the correct fix or not? > > > Ack. Fwiw, I believe that the reasoning put by Jani is perfeclty > reasonable, but it'll be great to hear others as well. Yes me to. However I don't think anything in this patch is at odds with the documentation Jani has referenced. regards, Peter.
Hi Peter, On 6 October 2016 at 11:48, Peter Griffin <peter.griffin@linaro.org> wrote: > Hi Emil, > > On Wed, 21 Sep 2016, Emil Velikov wrote: > >> On 20 September 2016 at 09:32, Peter Griffin <peter.griffin@linaro.org> wrote: >> > Hi Emil, >> > >> > On Tue, 20 Sep 2016, Emil Velikov wrote: >> > >> >> On 5 September 2016 at 14:16, Peter Griffin <peter.griffin@linaro.org> wrote: >> >> > ST_SLIM_REMOTEPROC must select REMOTEPROC, which exposes the following >> >> > recursive dependency. >> > >> > >> >> > >> >> From a humble skim through remoteproc, drm and a few other subsystems >> >> I think the above is wrong. All the drivers (outside of remoteproc), >> >> that I've seen, depend on the core component, they don't select it. >> > >> > I will let Bjorn comment on the remoteproc subsystem Kconfig design, and >> > why it is like it is. >> > >> > For this particular SLIM_RPROC I have added it to Kconfig in keeping with all >> > the other drivers in the remoteproc subsystem which has exposed this recursive >> > dependency issue. >> > >> > For this particular kconfig symbol a quick grep reveals more drivers in >> > the kernel using 'select', than 'depend on' >> > >> > git grep "select VIRTIO" | wc -l >> > 14 >> > >> > git grep "depends on VIRTIO" | wc -l >> > 10 >> > >> Might be worth taking a closer look into these at some point. > > VIRTIO has no dependencies, and is a non visible symbol. Its Kconfig help also > mentions that drivers should select it. > This is a (un)fortunate detail cannot/should not overrule the other arguments I've mentioned, should it ? >> >> > >> >> Furthermore most places explicitly hide the drivers from the menu if >> >> the core component isn't enabled. >> > >> > Remoteproc subsystem takes a different approach, the core code is only enabled >> > if a driver which relies on it is enabled. This IMHO makes sense, as >> > remoteproc is not widely used (only a few particular ARM SoC's). >> > >> > It is true that for subsystems which rely on the core component being >> > explicitly enabled, they often tend to hide drivers which depend on it >> > from the menu unless it is. This also makes sense. >> > >> >> >> >> Is there something that requires such a different/unusual behaviour in >> >> remoteproc ? >> >> >> > >> > I'm not sure it is that unusual...looking at config USB, it selects USB_COMMON in >> > mfd subsystem, client drivers select MFD_CORE. >> > >> On the USB case I'm not sure what the reasoning behind the USB vs >> USB_COMMON split. In seems that one could just fold them, but that's >> another topic. On the MFD side... it follows the select {,mis,ab}use. >> With one (the only one?) MFD driver not using/selecting MFD_CORE doing >> it's own version of mfd_add_devices... which could be reworked, >> possibly. > > Much like selecting VIRTIO in this patch, MFD_CORE is a non visible symbol > with no dependencies so it matches the documentation Jani referenced. > > I personally think the approach taken makes sense, as why would you want to have > a CONFIG_MFD_CORE=y symbol being enabled unless you actually have a MFD device > which uses it also enabled in your kernel? > > It seems to me a good solution to make the client drivers select the core > component so that it only gets enabled if at least one driver requires it. > This avoids having useless core code which will never be used compiled into the > kernel and in the end a smaller kernel size for a given kernel configuration (better > cache use etc etc). > >> > I've added Arnd to this thread, as I've seen lots of Kconfig patches from him >> > recently, maybe he has some thoughts on whether this is the correct fix or not? >> > >> Ack. Fwiw, I believe that the reasoning put by Jani is perfeclty >> reasonable, but it'll be great to hear others as well. > > Yes me to. However I don't think anything in this patch is at odds with the > documentation Jani has referenced. > It case it's not clear, let me elaborate: Yes, using depend might not be the most user-friendly thing to do and in the long term we might want to move to select. Yes, I agree with the argument about symbol visibility but that is not the only contributing factor. If one wants to pick on specific users which opt for $driver select $core they must do the same for $driver depends on $core. Check the number 'in favour" of each case and draw their conclusions ;-) In particular: both MFD_CORE and USB_COMMON, are _optional_ as only some drivers depends on them. Thus giving them as an example is wrong/irrelevant, I'm afraid. Thanks Emil
Hi Emil, On Thu, 06 Oct 2016, Emil Velikov wrote: > On 6 October 2016 at 10:37, Peter Griffin <peter.griffin@linaro.org> wrote: > > > In fact the help text for VIRTIO even states this option should be selected > > by any driver which implements virtio. > > > Almost but not quite. It says: > > "This option is selected by any driver which implements the virtio _bus_" > Ah I thought DRM_VIRTIO_GPU was implementing a virtio bus, bus it seems that it uses pci. Which does raise the question of why it is depending on VIRTIO at all and not VIRTIO_PCI. > REMOTEPROC obviously does that while the ST SLIM driver does not. Thus > the latter should _not_ select, be that explicitly or implicitly via > REMOTEPROC, the symbol. Yep OK. > > >> > >> People tend to abuse select because it's "convenient". If you depend, > >> but some of your dependencies aren't met, you're in for some digging > >> through Kconfig to find the missing deps. Just to make the option you > >> want visible in menuconfig. If you instead select something with > >> dependencies, it'll be right most of the time, and it's "convenient", > >> until it breaks. (And hey, it usually breaks for someone else with some > >> other config, so it's still convenient for you.) > > > > I'm sure they do but in this case it is actually the use of 'depends on' > > which has caused the breakage and inconvenience for somebody else and sadly this > > inconvienice is still on-going due to this patch not being applied or getting an > > acked-by from the appropriate maintainers. > > > Surely you're not saying that pre-existing driver following the > documentation, is 'causing breakage' for a new driver {ab,mis}using a > feature ? Your right I wasn't saying that :) My point was that this patch wasn't 'wrong' when referring to the Kconfig documentation Jani referenced as VIRTIO has no dependencies. Also I thought DRM_VIRTIO_GPU driver implemented a VIRTIO bus which re-enforced the view that it should be selecting VIRTIO. > > This reminds me an old saying: "If the shoe doesn’t fit, it doesn’t > mean there is anything wrong with your feet." If the shoe doesn't fit, chop off the leg :) > You seem to be suggesting the opposite ? > > >> > >> Perhaps kconfig should complain about selecting visible symbols and > >> symbols with dependencies. > > > > That sounds like it would be a useful addition. > > > > Is it possible to get this patch applied or an acked-by to avoid further delay > > to the fdma series? > > > Please don't apply duct tape, especially where it's _not_ needed. > > $ sed -i s/select REMOTEPROC/depends on REMOTEPROC/ drivers/remoteproc/Kconfig > > ... will resolve things in the right place. The alternative will lead > to random issues in other subsystems. > If Bjorn is OK with it, then it is fine with me. I will update remoteproc Kconfig setup in fdma v10, this will drop the requirement for this patch in drm subsystem. I can then send the whitespace cleanup patch separately to DRM ML. regards, Peter.
diff --git a/drivers/gpu/drm/virtio/Kconfig b/drivers/gpu/drm/virtio/Kconfig index e1afc3d..90357d9 100644 --- a/drivers/gpu/drm/virtio/Kconfig +++ b/drivers/gpu/drm/virtio/Kconfig @@ -1,6 +1,7 @@ config DRM_VIRTIO_GPU tristate "Virtio GPU driver" - depends on DRM && VIRTIO + depends on DRM + select VIRTIO select DRM_KMS_HELPER select DRM_TTM help
ST_SLIM_REMOTEPROC must select REMOTEPROC, which exposes the following recursive dependency. [..] drivers/video/fbdev/Kconfig:5:error: recursive dependency detected! For a resolution refer to Documentation/kbuild/kconfig-language.txt subsection "Kconfig recursive dependency limitations" drivers/video/fbdev/Kconfig:5: symbol FB is selected by DRM_KMS_FB_HELPER For a resolution refer to Documentation/kbuild/kconfig-language.txt subsection "Kconfig recursive dependency limitations" drivers/gpu/drm/Kconfig:42: symbol DRM_KMS_FB_HELPER depends on DRM_KMS_HELPER For a resolution refer to Documentation/kbuild/kconfig-language.txt subsection "Kconfig recursive dependency limitations" drivers/gpu/drm/Kconfig:36: symbol DRM_KMS_HELPER is selected by DRM_VIRTIO_GPU For a resolution refer to Documentation/kbuild/kconfig-language.txt subsection "Kconfig recursive dependency limitations" drivers/gpu/drm/virtio/Kconfig:1: symbol DRM_VIRTIO_GPU depends on VIRTIO For a resolution refer to Documentation/kbuild/kconfig-language.txt subsection "Kconfig recursive dependency limitations" drivers/virtio/Kconfig:1: symbol VIRTIO is selected by REMOTEPROC For a resolution refer to Documentation/kbuild/kconfig-language.txt subsection "Kconfig recursive dependency limitations" drivers/remoteproc/Kconfig:4: symbol REMOTEPROC is selected by ST_SLIM_REMOTEPROC For a resolution refer to Documentation/kbuild/kconfig-language.txt subsection "Kconfig recursive dependency limitations" drivers/remoteproc/Kconfig:103: symbol ST_SLIM_REMOTEPROC is selected by ST_FDMA For a resolution refer to Documentation/kbuild/kconfig-language.txt subsection "Kconfig recursive dependency limitations" drivers/dma/Kconfig:440: symbol ST_FDMA depends on DMADEVICES For a resolution refer to Documentation/kbuild/kconfig-language.txt subsection "Kconfig recursive dependency limitations" drivers/dma/Kconfig:5: symbol DMADEVICES is selected by SND_SOC_SH4_SIU For a resolution refer to Documentation/kbuild/kconfig-language.txt subsection "Kconfig recursive dependency limitations" sound/soc/sh/Kconfig:29: symbol SND_SOC_SH4_SIU is selected by SND_SIU_MIGOR For a resolution refer to Documentation/kbuild/kconfig-language.txt subsection "Kconfig recursive dependency limitations" sound/soc/sh/Kconfig:64: symbol SND_SIU_MIGOR depends on I2C For a resolution refer to Documentation/kbuild/kconfig-language.txt subsection "Kconfig recursive dependency limitations" drivers/i2c/Kconfig:7: symbol I2C is selected by FB_DDC For a resolution refer to Documentation/kbuild/kconfig-language.txt subsection "Kconfig recursive dependency limitations" drivers/video/fbdev/Kconfig:63: symbol FB_DDC is selected by FB_CYBER2000_DDC For a resolution refer to Documentation/kbuild/kconfig-language.txt subsection "Kconfig recursive dependency limitations" drivers/video/fbdev/Kconfig:378: symbol FB_CYBER2000_DDC depends on FB_CYBER2000 For a resolution refer to Documentation/kbuild/kconfig-language.txt subsection "Kconfig recursive dependency limitations" drivers/video/fbdev/Kconfig:366: symbol FB_CYBER2000 depends on FB which is due to drivers/gpu/drm/virtio/Kconfig depending on VIRTIO. Fix by dropping depend and switching to select VIRTIO. Signed-off-by: Peter Griffin <peter.griffin@linaro.org> --- drivers/gpu/drm/virtio/Kconfig | 3 ++- 1 file changed, 2 insertions(+), 1 deletion(-)