drm: fb_helper: fix CONFIG_FB dependency
Subsystems (2)
PATCH RESEND 1/1 drm: fb_helper: fix CONFIG_FB dependency
From: Arnd Bergmann <[email protected]>
With CONFIG_FB=m and CONFIG_DRM=y, we get a link error in the fb helper:
aarch64-linux-ld: drivers/gpu/drm/drm_fb_helper.o: in function `drm_fb_helper_alloc_fbi':
(.text+0x10cc): undefined reference to `framebuffer_alloc'
Tighten the dependency so it is only allowed in the case that DRM can link against FB.
---
I posted this in July, the patch is still required and should work on its own.
--- drivers/gpu/drm/Kconfig | 2 +- 1 file changed, 1 insertion(+), 1 deletion(-)
drivers/gpu/drm/Kconfig +1 -1
@@ -103,7 +103,7 @@ config DRM_DEBUG_DP_MST_TOPOLOGY_REFS config DRM_FBDEV_EMULATION bool "Enable legacy fbdev support for your modesetting driver" depends on DRM - depends on FB + depends on FB=y || FB=DRM select DRM_KMS_HELPER select FB_CFB_FILLRECT select FB_CFB_COPYAREA
Signature · 1 line
2.29.2
Quoted 7 lines · Arnd Bergmann
From: Arnd Bergmann <[email protected]>
With
CONFIG_FB=mandCONFIG_DRM=y, we get a link error in the fb helper:aarch64-linux-ld:
drivers/gpu/drm/drm_fb_helper.o: in function `drm_fb_helper_alloc_fbi':
(.text+0x10cc): undefined reference to `framebuffer_alloc'
Quoted 6 lines
Tighten the dependency so it is only allowed in the case that DRM can link against FB.
Fixes f611b1e7624c ("drm: Avoid circular dependencies for CONFIG_FB")Signed-off-by Arnd Bergmann <[email protected]>
Thanks for fixing this!
Signature · 1 line
Kees Cook
Quoted 11 lines · Kees Cook
Quoted 10 lines · Arnd Bergmann
From: Arnd Bergmann <[email protected]>
With
CONFIG_FB=mandCONFIG_DRM=y, we get a link error in the fb helper:aarch64-linux-ld:
drivers/gpu/drm/drm_fb_helper.o: in function `drm_fb_helper_alloc_fbi':
(.text+0x10cc): undefined reference to `framebuffer_alloc'Tighten the dependency so it is only allowed in the case that DRM can link against FB.
Quoted 7 lines
Quoted 3 lines
Fixes f611b1e7624c ("drm: Avoid circular dependencies for CONFIG_FB")Signed-off-by Arnd Bergmann <[email protected]>Thanks for fixing this!
Reviewed-by Kees Cook <[email protected]>
Stuffed into drm-misc-next.
Arnd, I guess I still can't volunteer you for commit rights so I don't have to bother with this anymore? It's nice to be lazy and comfy :-)
Cheers, Daniel
Signature · 3 lines
Daniel Vetter Software Engineer, Intel Corporation http://blog.ffwll.ch
Quoted 16 lines · Daniel Vetter
Quoted 15 lines · Kees Cook
Quoted 13 lines · Arnd Bergmann
From: Arnd Bergmann <[email protected]>
With
CONFIG_FB=mandCONFIG_DRM=y, we get a link error in the fb helper:aarch64-linux-ld:
drivers/gpu/drm/drm_fb_helper.o: in function `drm_fb_helper_alloc_fbi':
(.text+0x10cc): undefined reference to `framebuffer_alloc'Tighten the dependency so it is only allowed in the case that DRM can link against FB.
Fixes f611b1e7624c ("drm: Avoid circular dependencies for CONFIG_FB")Signed-off-by Arnd Bergmann <[email protected]>
Quoted 5 lines
Quoted 3 lines
Thanks for fixing this!
Reviewed-by Kees Cook <[email protected]>Stuffed into drm-misc-next.
The problem is, I don't think the patch is semantically correct.
drm_fb_helper.o is not part of drm.ko, it's part of drm_kms_helper.ko. This adds some sort of indirect dependency via DRM which might work, maybe by coincidence, maybe not - but it's certainly not obvious.
The likely culprit is, again, the overuse of select, and in this case select DRM_KMS_HELPER. And DRM_KMS_HELPER should depend on FB if DRM_FBDEV_EMULATION=y. That's the problem.
All of the drm Kconfigs could use an overhaul to be semantically correct, but that's a hill nobody wants to die on. Instead we keep piling up tweaks to paper over the issues, ad infinitum.
(And this ties to a previous comment I had about the organization of files under drm/, a hundred files in one big lump that belong to different modules, and it's not helping people figure out the dependencies.)
BR,
Jani.
PS. I was brought here via [1] which is another complicated "fix" to the same problem.
[1] https://lore.kernel.org/r/[email protected]
Signature · 1 line
Jani Nikula, Intel Open Source Graphics Center
Quoted 25 lines · Jani Nikula
Quoted 16 lines · Daniel Vetter
Quoted 15 lines · Kees Cook
Quoted 13 lines · Arnd Bergmann
From: Arnd Bergmann <[email protected]>
With
CONFIG_FB=mandCONFIG_DRM=y, we get a link error in the fb helper:aarch64-linux-ld:
drivers/gpu/drm/drm_fb_helper.o: in function `drm_fb_helper_alloc_fbi':
(.text+0x10cc): undefined reference to `framebuffer_alloc'Tighten the dependency so it is only allowed in the case that DRM can link against FB.
Fixes f611b1e7624c ("drm: Avoid circular dependencies for CONFIG_FB")Signed-off-by Arnd Bergmann <[email protected]>Quoted 5 lines
Quoted 3 lines
Thanks for fixing this!
Reviewed-by Kees Cook <[email protected]>Stuffed into drm-misc-next.
The problem is, I don't think the patch is semantically correct.
Quoted 8 lines
drm_fb_helper.o is not part of drm.ko, it's part of drm_kms_helper.ko. This adds some sort of indirect dependency via DRM which might work, maybe by coincidence, maybe not - but it's certainly not obvious.
The likely culprit is, again, the overuse of select, and in this case select
DRM_KMS_HELPER. AndDRM_KMS_HELPERshould depend on FB if DRM_FBDEV_EMULATION=y. That's the problem.
Almost all of the recurring Kconfig related dependency issues would go away by following Documentation/kbuild/kconfig-language.rst:
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.
If the kconfig parser had a lint mode to issue a warning for all of those uses, and someone persistent enough followed through with fixing them, we'd all be better off.
Oh, and maybe the menuconfig tools also need better ways to recursively enable config options with dependencies, because one of the reasons people like select is the convenience of just enabling a config option, and it selects everything that's needed (albeit with the occasional dependency issues). With dependencies, you need to start with the leaf dependencies and work your way up to what you need, and it's not easy.
BR,
Jani.
Quoted 20 lines
All of the drm Kconfigs could use an overhaul to be semantically correct, but that's a hill nobody wants to die on. Instead we keep piling up tweaks to paper over the issues, ad infinitum.
(And this ties to a previous comment I had about the organization of files under drm/, a hundred files in one big lump that belong to different modules, and it's not helping people figure out the dependencies.)
BR,
Jani.PS. I was brought here via [1] which is another complicated "fix" to the same problem.
Signature · 1 line
Jani Nikula, Intel Open Source Graphics Center
Quoted 21 lines · Jani Nikula
Quoted 20 lines · Daniel Vetter
Quoted 18 lines · Kees Cook
Quoted 7 lines · Arnd Bergmann
From: Arnd Bergmann <[email protected]>
With
CONFIG_FB=mandCONFIG_DRM=y, we get a link error in the fb helper:aarch64-linux-ld:
drivers/gpu/drm/drm_fb_helper.o: in function `drm_fb_helper_alloc_fbi':
(.text+0x10cc): undefined reference to `framebuffer_alloc'Quoted 6 lines
Tighten the dependency so it is only allowed in the case that DRM can link against FB.
Fixes f611b1e7624c ("drm: Avoid circular dependencies for CONFIG_FB")Signed-off-by Arnd Bergmann <[email protected]>Thanks for fixing this!
Reviewed-by Kees Cook <[email protected]>
Quoted 8 lines
Quoted 1 line
Stuffed into drm-misc-next.
The problem is, I don't think the patch is semantically correct.
drm_fb_helper.o is not part of drm.ko, it's part of drm_kms_helper.ko. This adds some sort of indirect dependency via DRM which might work, maybe by coincidence, maybe not - but it's certainly not obvious.
Right, how about this change on top?
drivers/gpu/drm/Kconfig +2 -3
@@ -117,9 +117,8 @@ config DRM_DEBUG_MODESET_LOCK config DRM_FBDEV_EMULATION bool "Enable legacy fbdev support for your modesetting driver" - depends on DRM - depends on FB=y || FB=DRM - select DRM_KMS_HELPER + depends on DRM_KMS_HELPER + depends on FB=y || FB=DRM_KMS_HELPER select FB_CFB_FILLRECT select FB_CFB_COPYAREA select FB_CFB_IMAGEBLIT
That would probably make it work for DRM=y, FB=m, DRM_KMS_HELPER=m, but it needs more randconfig testing, which I can help with.
Quoted 3 lines
The likely culprit is, again, the overuse of select, and in this case select
DRM_KMS_HELPER. AndDRM_KMS_HELPERshould depend on FB if DRM_FBDEV_EMULATION=y. That's the problem.
This is something we can't easily express in Kconfig, as we can't add the dependency to a symbol that only gets selected by other drivers, which is why the dependency has to be in the user-visible symbol, in this case DRM_FBDEV_EMULATION.
Quoted 3 lines
All of the drm Kconfigs could use an overhaul to be semantically correct, but that's a hill nobody wants to die on. Instead we keep piling up tweaks to paper over the issues, ad infinitum.
Yes, that is a big issue, though we have similar problems with drivers/media and net/.
On a related note, I did manage to sort out the backlight dependency issue (intel_panel.c:(.text+0x2f58): undefined reference to `backlight_device_register'),
but haven't sent that one again yet, but I can if you like. This one changes DRM_I915 and all of drivers/video/fbdev from 'select BACKLIGHT_CLASS_DEVICE' to 'depends on', which I think moves everything into broadly the right direction.
Let me know if you would like me to send those now, or have a look at the top 3 patches in [1] if you are interested. This has passed a few thousand randconfig
builds and should not depend on additional patches.
Arnd
[1] https://git.kernel.org/pub/scm/linux/kernel/git/arnd/playground.git/log/?h=randconfig-5.16-next
Quoted 1 line · Arnd Bergmann
On Wed, Oct 27, 2021 at 1:47 PM Jani Nikula <[email protected]> wrote:
[snip]
Quoted 4 lines
Quoted 4 lines
drm_fb_helper.o is not part of drm.ko, it's part of drm_kms_helper.ko. This adds some sort of indirect dependency via DRM which might work, maybe by coincidence, maybe not - but it's certainly not obvious.
Indeed, you are correct that's not semantically correct.
Quoted 21 lines
Right, how about this change on top?
drivers/gpu/drm/Kconfig +2
-3@@ -117,9 +117,8 @@ config DRM_DEBUG_MODESET_LOCK config DRM_FBDEV_EMULATION bool "Enable legacy fbdev support for your modesetting driver" - depends on DRM - depends on FB=y || FB=DRM - select DRM_KMS_HELPER + depends on DRM_KMS_HELPER + depends on FB=y || FB=DRM_KMS_HELPER select FB_CFB_FILLRECT select FB_CFB_COPYAREA select FB_CFB_IMAGEBLITThat would probably make it work for DRM=y, FB=m, DRM_KMS_HELPER=m, but it needs more randconfig testing, which I can help with.
Quoted 9 lines
Quoted 3 lines
The likely culprit is, again, the overuse of select, and in this case select
DRM_KMS_HELPER. AndDRM_KMS_HELPERshould depend on FB if DRM_FBDEV_EMULATION=y. That's the problem.This is something we can't easily express in
Kconfig, as we can't add the dependency to a symbol that only gets selected by other drivers, which is why the dependency has to be in the user-visible symbol, in this caseDRM_FBDEV_EMULATION.
Why the dependency has to be in a user-visible symbol? What could be the problem with having something like:
drivers/gpu/drm/Kconfig +1 -1
@@ -82,6 +82,7 @@ config DRM_DEBUG_SELFTEST config DRM_KMS_HELPER tristate depends on DRM + depends on (DRM_FBDEV_EMULATION && FB) || !DRM_FBDEV_EMULATION help CRTC helpers for KMS drivers.
@@ -104,7 +105,6 @@ config DRM_FBDEV_EMULATION bool "Enable legacy fbdev support for your modesetting driver" depends on DRM depends on FB - select DRM_KMS_HELPER select FB_CFB_FILLRECT select FB_CFB_COPYAREA select FB_CFB_IMAGEBLIT
Best regards,
Signature · 3 lines
Javier Martinez Canillas Linux Engineering Red Hat
Quoted 23 lines · Javier Martinez Canillas
Quoted 6 lines
This is something we can't easily express in
Kconfig, as we can't add the dependency to a symbol that only gets selected by other drivers, which is why the dependency has to be in the user-visible symbol, in this caseDRM_FBDEV_EMULATION.Why the dependency has to be in a user-visible symbol? What could be the problem with having something like:
drivers/gpu/drm/Kconfig +1
-0@@ -82,6 +82,7 @@ config DRM_DEBUG_SELFTEST config DRM_KMS_HELPER tristate depends on DRM + depends on (DRM_FBDEV_EMULATION && FB) || !DRM_FBDEV_EMULATION help CRTC helpers for KMS drivers.
drivers/gpu/drm/Kconfig config DRM_FBDEV_EMULATION · line 110
Quoted 7 lines
drivers/gpu/drm/Kconfig +0
-1@@ -104,7 +105,6 @@ config DRM_FBDEV_EMULATION bool "Enable legacy fbdev support for your modesetting driver" depends on DRM depends on FB - select DRM_KMS_HELPER select FB_CFB_FILLRECT select FB_CFB_COPYAREA select FB_CFB_IMAGEBLIT
This fails because of all the other drivers that try to 'select DRM_KMS_HELPER'. Kconfig will now complain about a symbol that gets selected while its dependencies
are not met.
To work around that, every single driver that has 'selects DRM_KMS_HELPER' would now have to also list 'depends on (DRM_FBDEV_EMULATION && FB) || !DRM_FBDEV_EMULATION'.
Arnd
Quoted 51 lines · Javier Martinez Canillas
Quoted 1 line · Arnd Bergmann
On Wed, Oct 27, 2021 at 1:47 PM Jani Nikula <[email protected]> wrote:
[snip]
Quoted 4 lines
Quoted 4 lines
drm_fb_helper.o is not part of drm.ko, it's part of drm_kms_helper.ko. This adds some sort of indirect dependency via DRM which might work, maybe by coincidence, maybe not - but it's certainly not obvious.
Indeed, you are correct that's not semantically correct.
Quoted 21 lines
Right, how about this change on top?
drivers/gpu/drm/Kconfig +2
-3@@ -117,9 +117,8 @@ config DRM_DEBUG_MODESET_LOCK config DRM_FBDEV_EMULATION bool "Enable legacy fbdev support for your modesetting driver" - depends on DRM - depends on FB=y || FB=DRM - select DRM_KMS_HELPER + depends on DRM_KMS_HELPER + depends on FB=y || FB=DRM_KMS_HELPER select FB_CFB_FILLRECT select FB_CFB_COPYAREA select FB_CFB_IMAGEBLITThat would probably make it work for DRM=y, FB=m, DRM_KMS_HELPER=m, but it needs more randconfig testing, which I can help with.
Quoted 9 lines
Quoted 3 lines
The likely culprit is, again, the overuse of select, and in this case select
DRM_KMS_HELPER. AndDRM_KMS_HELPERshould depend on FB if DRM_FBDEV_EMULATION=y. That's the problem.This is something we can't easily express in
Kconfig, as we can't add the dependency to a symbol that only gets selected by other drivers, which is why the dependency has to be in the user-visible symbol, in this caseDRM_FBDEV_EMULATION.Why the dependency has to be in a user-visible symbol? What could be the problem with having something like:
drivers/gpu/drm/Kconfig +0
-0
drivers/gpu/drm/Kconfig config DRM_DEBUG_SELFTEST · line 85
Quoted 4 lines
drivers/gpu/drm/Kconfig +1
-0@@ -82,6 +82,7 @@ config DRM_DEBUG_SELFTEST config DRM_KMS_HELPER tristate depends on DRM + depends on (DRM_FBDEV_EMULATION && FB) || !DRM_FBDEV_EMULATION
To me, this seems like the right solution. Depend on FB if DRM_FBDEV_EMULATION is enabled. That's exactly what the relationship is.
BR,
Jani.
Quoted 13 lines
drivers/gpu/drm/Kconfig +0
-1help CRTC helpers for KMS drivers.@@ -104,7 +105,6 @@ config DRM_FBDEV_EMULATION bool "Enable legacy fbdev support for your modesetting driver" depends on DRM depends on FB - select DRM_KMS_HELPER select FB_CFB_FILLRECT select FB_CFB_COPYAREA select FB_CFB_IMAGEBLITBest regards,
Signature · 1 line
Jani Nikula, Intel Open Source Graphics Center
On 10/27/21 14:52, Arnd Bergmann wrote:
[snip]
Quoted 15 lines
drivers/gpu/drm/Kconfig config DRM_FBDEV_EMULATION · line 110
Quoted 9 lines
drivers/gpu/drm/Kconfig +0
-1@@ -104,7 +105,6 @@ config DRM_FBDEV_EMULATION bool "Enable legacy fbdev support for your modesetting driver" depends on DRM depends on FB - select DRM_KMS_HELPER select FB_CFB_FILLRECT select FB_CFB_COPYAREA select FB_CFB_IMAGEBLITThis fails because of all the other drivers that try to 'select
DRM_KMS_HELPER'.Kconfigwill now complain about a symbol that gets selected while its dependencies
are not met.
Quoted 4 lines
To work around that, every single driver that has 'selects
DRM_KMS_HELPER' would now have to also list 'depends on (DRM_FBDEV_EMULATION&& FB) || !DRM_FBDEV_EMULATION'.
Ah, I see now. Thanks a lot for the explanation.
Quoted 2 lines
Arnd
Best regards,
Signature · 3 lines
Javier Martinez Canillas Linux Engineering Red Hat
Jani Nikula
Acked-byQuoted 8 lines · Arnd Bergmann
On a related note, I did manage to sort out the backlight dependency issue (intel_panel.c:(.text+0x2f58): undefined reference to `backlight_device_register'),
but haven't sent that one again yet, but I can if you like. This one changesDRM_I915and all ofdrivers/video/fbdevfrom 'selectBACKLIGHT_CLASS_DEVICE' to 'depends on', which I think moves everything into broadly the right direction.
Quoted 4 lines
Let me know if you would like me to send those now, or have a look at the top 3 patches in [1] if you are interested. This has passed a few thousand randconfig
builds and should not depend on additional patches.
FWIW,
on the patches. I think I've sent patches before to do the same change from "select" to "depends on", but they went nowhere. IIRC the opposition was that people wanted to be able to find and enable their driver in menuconfig without first having to enable BACKLIGHT_CLASS_DEVICE.
BR,
Jani.
Quoted 4 lines
Arnd
[1] https://git.kernel.org/pub/scm/linux/kernel/git/arnd/playground.git/log/?h=randconfig-5.16-next
Signature · 1 line
Jani Nikula, Intel Open Source Graphics Center
Quoted 38 lines · Arnd Bergmann
Quoted 23 lines · Javier Martinez Canillas
Quoted 6 lines
This is something we can't easily express in
Kconfig, as we can't add the dependency to a symbol that only gets selected by other drivers, which is why the dependency has to be in the user-visible symbol, in this caseDRM_FBDEV_EMULATION.Why the dependency has to be in a user-visible symbol? What could be the problem with having something like:
drivers/gpu/drm/Kconfig +1
-0@@ -82,6 +82,7 @@ config DRM_DEBUG_SELFTEST config DRM_KMS_HELPER tristate depends on DRM + depends on (DRM_FBDEV_EMULATION && FB) || !DRM_FBDEV_EMULATION help CRTC helpers for KMS drivers.drivers/gpu/drm/Kconfig config DRM_FBDEV_EMULATION · line 110
Quoted 7 lines
drivers/gpu/drm/Kconfig +0
-1@@ -104,7 +105,6 @@ config DRM_FBDEV_EMULATION bool "Enable legacy fbdev support for your modesetting driver" depends on DRM depends on FB - select DRM_KMS_HELPER select FB_CFB_FILLRECT select FB_CFB_COPYAREA select FB_CFB_IMAGEBLITThis fails because of all the other drivers that try to 'select
DRM_KMS_HELPER'.Kconfigwill now complain about a symbol that gets selected while its dependencies
are not met.
Quoted 3 lines
To work around that, every single driver that has 'selects
DRM_KMS_HELPER' would now have to also list 'depends on (DRM_FBDEV_EMULATION&& FB) || !DRM_FBDEV_EMULATION'.
So the fix would be that nobody selects DRM_KMS_HELPER...
BR,
Jani.
Quoted 2 lines
Arnd
Signature · 1 line
Jani Nikula, Intel Open Source Graphics Center
On 10/27/21 14:55, Jani Nikula wrote:
[snip]
Quoted 16 lines
drivers/gpu/drm/Kconfig config DRM_DEBUG_SELFTEST · line 85
Quoted 12 lines
Why the dependency has to be in a user-visible symbol? What could be the problem with having something like:
drivers/gpu/drm/Kconfig +1
-0@@ -82,6 +82,7 @@ config DRM_DEBUG_SELFTEST config DRM_KMS_HELPER tristate depends on DRM + depends on (DRM_FBDEV_EMULATION && FB) || !DRM_FBDEV_EMULATIONTo me, this seems like the right solution. Depend on FB if
DRM_FBDEV_EMULATIONis enabled. That's exactly what the relationship is.
The problem as Arnd explained is that then this relationship will have to be expressed in all the Kconfig symbols that select DRM_KMS_HELPER.
Otherwise the symbol will happily select the wrong state and even when a warning is printed by Kconfig, it will just set an invalid configuration.
For example with CONFIG_FB=m (that led to the linker errors if the symbol is also not CONFIG_DRM_KMS_HELPER=m) and CONFIG_SIMPLEDRM=y (that selects CONFIG_DRM_KMS_HELPER), this would cause the following unmet dependencies:
$ make prepare modules_prepare WARNING: unmet direct dependencies detected for DRM_KMS_HELPER Depends on [m]: HAS_IOMEM [=y] && DRM [=y] && (DRM_FBDEV_EMULATION [=y] && FB [=m] || !DRM_FBDEV_EMULATION [=y]) Selected by [y]: - DRM_SIMPLEDRM [=y] && HAS_IOMEM [=y] && DRM [=y] Selected by [m]: - DRM_I915 [=m] && HAS_IOMEM [=y] && DRM [=y] && X86 [=y] && PCI [=y] - DRM_VIRTIO_GPU [=m] && HAS_IOMEM [=y] && DRM [=y] && VIRTIO_MENU [=y] && MMU [=y]
so CONFIG_DRM_KMS_HELPER will wrongly set to =y which will cause the issue.
Best regards,
Signature · 3 lines
Javier Martinez Canillas Linux Engineering Red Hat
Javier Martinez Canillas
Reviewed-byOn 10/27/21 14:18, Arnd Bergmann wrote:
[snip]
Quoted 17 lines
Right, how about this change on top?
drivers/gpu/drm/Kconfig +2
-3@@ -117,9 +117,8 @@ config DRM_DEBUG_MODESET_LOCK config DRM_FBDEV_EMULATION bool "Enable legacy fbdev support for your modesetting driver" - depends on DRM - depends on FB=y || FB=DRM - select DRM_KMS_HELPER + depends on DRM_KMS_HELPER + depends on FB=y || FB=DRM_KMS_HELPER select FB_CFB_FILLRECT select FB_CFB_COPYAREA select FB_CFB_IMAGEBLIT
Quoted 2 lines
That would probably make it work for DRM=y, FB=m, DRM_KMS_HELPER=m, but it needs more randconfig testing, which I can help with.
Looks good to me as well.
Best regards,
Signature · 3 lines
Javier Martinez Canillas Linux Engineering Red Hat
Quoted 11 lines · Jani Nikula
Quoted 8 lines · Arnd Bergmann
This fails because of all the other drivers that try to 'select
DRM_KMS_HELPER'.Kconfigwill now complain about a symbol that gets selected while its dependencies
are not met.To work around that, every single driver that has 'selects
DRM_KMS_HELPER' would now have to also list 'depends on (DRM_FBDEV_EMULATION&& FB) || !DRM_FBDEV_EMULATION'.So the fix would be that nobody selects
DRM_KMS_HELPER...
That's not going to help in this case, the way the helper functions work is that you select them as needed, and you avoid the other dependencies. This part works fine.
We could probably get rid of this symbol by just making it an unconditional part of drm.ko, as almost every driver ends up using it anyway.
Arguably, this would make the end result worse, as you'd again get drm.ko itself to link against the old framebuffer code.
What I'm not sure about is whether drivers/video/fbdev/core/fb.ko could be split up into smaller parts so DRM_FBDEV_EMULATION could only depend on a set of common code without the bits that are needed for the classic fbdev drivers.
Arnd
On 10/27/21 15:25, Arnd Bergmann wrote:
[snip]
Quoted 7 lines
That's not going to help in this case, the way the helper functions work is that you select them as needed, and you avoid the other dependencies. This part works fine.
We could probably get rid of this symbol by just making it an unconditional part of drm.ko, as almost every driver ends up using it anyway.
Quoted 8 lines
Arguably, this would make the end result worse, as you'd again get drm.ko itself to link against the old framebuffer code.
What I'm not sure about is whether
drivers/video/fbdev/core/fb.kocould be split up into smaller parts soDRM_FBDEV_EMULATIONcould only depend on a set of common code without the bits that are needed for the classic fbdev drivers.
I attempted to do something like that but the changes were nacked:
https://patchwork.kernel.org/project/linux-fbdev/list/?series=538227
Quoted 2 lines
Arnd
Best regards,
Signature · 3 lines
Javier Martinez Canillas Linux Engineering Red Hat
Quoted 21 lines · Javier Martinez Canillas
On 10/27/21 14:18, Arnd Bergmann wrote:
[snip]
Quoted 17 lines
Right, how about this change on top?
drivers/gpu/drm/Kconfig +2
-3@@ -117,9 +117,8 @@ config DRM_DEBUG_MODESET_LOCK config DRM_FBDEV_EMULATION bool "Enable legacy fbdev support for your modesetting driver" - depends on DRM - depends on FB=y || FB=DRM - select DRM_KMS_HELPER + depends on DRM_KMS_HELPER + depends on FB=y || FB=DRM_KMS_HELPER select FB_CFB_FILLRECT select FB_CFB_COPYAREA select FB_CFB_IMAGEBLIT
Quoted 6 lines
Quoted 2 lines
That would probably make it work for DRM=y, FB=m, DRM_KMS_HELPER=m, but it needs more randconfig testing, which I can help with.
Looks good to me as well.
Reviewed-by Javier Martinez Canillas <[email protected]>
Is the mess I created sorted now, or something for me to do? I'm terribly burried in stuff :-/
-Daniel
Signature · 3 lines
Daniel Vetter Software Engineer, Intel Corporation http://blog.ffwll.ch
Quoted 28 lines · Daniel Vetter
Quoted 26 lines · Javier Martinez Canillas
On 10/27/21 14:18, Arnd Bergmann wrote:
[snip]
Quoted 17 lines
Right, how about this change on top?
drivers/gpu/drm/Kconfig +2
-3@@ -117,9 +117,8 @@ config DRM_DEBUG_MODESET_LOCK config DRM_FBDEV_EMULATION bool "Enable legacy fbdev support for your modesetting driver" - depends on DRM - depends on FB=y || FB=DRM - select DRM_KMS_HELPER + depends on DRM_KMS_HELPER + depends on FB=y || FB=DRM_KMS_HELPER select FB_CFB_FILLRECT select FB_CFB_COPYAREA select FB_CFB_IMAGEBLITQuoted 2 lines
That would probably make it work for DRM=y, FB=m, DRM_KMS_HELPER=m, but it needs more randconfig testing, which I can help with.
Looks good to me as well.
Quoted 4 lines
Quoted 1 line
Reviewed-by Javier Martinez Canillas <[email protected]>Is the mess I created sorted now, or something for me to do? I'm terribly burried in stuff :-/
I have done a few days worth of build testing with that patch applied, and did not see any other problems. I've written a proper description and submitted it as
https://lore.kernel.org/all/[email protected]/T
The version you have in linux-next works correctly, but this patch on top is an improvement.
Arnd