drm/simpledrm: Only advertise formats that are supported
Subsystems (3)
PATCH 1/1 drm/simpledrm: Only advertise formats that are supported
Until now, simpledrm unconditionally advertised all formats that can be supported natively as conversions. However, we don't actually have a full conversion matrix of helpers. Although the list is arguably provided to userspace in precedence order, userspace can pick something out-of-order (and thus break when it shouldn't), or simply only support a format that is unsupported (and thus think it can work, which results in the appearance of a hang as FB blits fail later on, instead of the initialization error you'd expect in this case).
Split up the format table into separate ones for each required subset, and then pick one based on the native format. Also remove the native<->conversion overlap check from the helper (which doesn't make sense any more, since the native format is advertised anyway and this way RGB565/RGB888 can share a format table), and instead print the same message in simpledrm when the native format is not one for which we have conversions at all.
This fixes a real user regression where the ?RGB2101010 support commit started advertising it unconditionally where not supported, and KWin decided to start to use it over the native format, but also the fixes the spurious RGB565/RGB888 formats which have been wrongly unconditionally advertised since the dawn of simpledrm.
Note: this patch is merged because splitting it into two patches, one for the helper and one for simpledrm, would regress at the midpoint regardless of the order. If simpledrm is changed first, that would break working conversions to RGB565/RGB888 (since those share a table that does not include the native formats). If the helper is changed first, it would start spuriously advertising all conversion formats when the native format doesn't have any supported conversions at all.
--- drivers/gpu/drm/drm_format_helper.c | 15 ------- drivers/gpu/drm/tiny/simpledrm.c | 62 +++++++++++++++++++++++++---- 2 files changed, 55 insertions(+), 22 deletions(-)
drivers/gpu/drm/drm_format_helper.c +0 -15
@@ -864,20 +864,6 @@ size_t drm_fb_build_fourcc_list(struct drm_device *dev, ++fourccs; } -/* -* The plane's atomic_update helper converts the framebuffer's color format -* to a native format when copying to device memory. -* -* If there is not a single format supported by both, device and -* driver, the native formats are likely not supported by the conversion -* helpers. Therefore *only* support the native formats and add a -* conversion helper ASAP. -*/ -if (!found_native) { -drm_warn (dev,"Format conversion helpers required to add extra formats.\n" ); - goto out; - } -/* * The extra formats, emulated by the driver, go second. */
@@ -898,7 +884,6 @@ size_t drm_fb_build_fourcc_list(struct drm_device *dev, ++fourccs; } -out:return fourccs - fourccs_out; }EXPORT_SYMBOL (drm_fb_build_fourcc_list);
drivers/gpu/drm/tiny/simpledrm.c +55 -7
94 lines not on this page · raw message
Signature · 1 line
2.35.1
Quoted 36 lines · Hector Martin
Until now, simpledrm unconditionally advertised all formats that can be supported natively as conversions. However, we don't actually have a full conversion matrix of helpers. Although the list is arguably provided to userspace in precedence order, userspace can pick something out-of-order (and thus break when it shouldn't), or simply only support a format that is unsupported (and thus think it can work, which results in the appearance of a hang as FB blits fail later on, instead of the initialization error you'd expect in this case).
Split up the format table into separate ones for each required subset, and then pick one based on the native format. Also remove the native<->conversion overlap check from the helper (which doesn't make sense any more, since the native format is advertised anyway and this way RGB565/RGB888 can share a format table), and instead print the same message in simpledrm when the native format is not one for which we have conversions at all.
This fixes a real user regression where the ?RGB2101010 support commit started advertising it unconditionally where not supported, and KWin decided to start to use it over the native format, but also the fixes the spurious RGB565/RGB888 formats which have been wrongly unconditionally advertised since the dawn of simpledrm.
Note: this patch is merged because splitting it into two patches, one for the helper and one for simpledrm, would regress at the midpoint regardless of the order. If simpledrm is changed first, that would break working conversions to RGB565/RGB888 (since those share a table that does not include the native formats). If the helper is changed first, it would start spuriously advertising all conversion formats when the native format doesn't have any supported conversions at all.
Acked-by Pekka Paalanen <[email protected]>Fixes 6ea966fca084 ("drm/simpledrm: Add [AX]RGB2101010 formats")Fixes 11e8f5fd223b ("drm: Add simpledrm driver")Signed-off-by Hector Martin <[email protected]>
Quoted 5 lines
--- drivers/gpu/drm/drm_format_helper.c | 15 ------- drivers/gpu/drm/tiny/simpledrm.c | 62 +++++++++++++++++++++++++---- 2 files changed, 55 insertions(+), 22 deletions(-)
To answer some issues that came up on IRC:
Q: Why not move this logic / the tables to the helper?
A: Because simpledrm is the only user so far, and this patch is Cc:
stable because we have an actual regression that broke KDE. I'm going for the minimal patch that keeps everything that worked to this day working, and stops advertising things that never worked, no more, no less. Future refactoring can always happen later (and is probably a good idea when other drivers start using the helper).
Q: XRGB8888 is supposed to be the only canonical format. Why not just drop everything but conversions to/from XRGB8888?
A: Because that would regress things that work today, and could break existing userspace on some platforms. That may be a good idea, but I think we should fix the bugs first, and leave the discussion of whether we want to actually remove existing functionality for later.
Q: Why not just add a conversion from XRGB2101010 to XRGB8888?
A: Because that would only fix KDE, and would make it slower vs. not advertising XRGB2101010 at all (double conversions, plus kernel conversion can be slower). Plus, it doesn't make any sense as it only fills in one entry in the conversion matrix. If we wanted to actually fill out the conversion matrix, and thus support everything simpledrm has advertised to day correctly, we would need helpers for:
rgb565->rgb888rgb888->rgb565rgb565->xrgb2101010rgb888->xrgb2101010xrgb2101010->rgb565xrgb2101010->rgb888xrgb2101010->xrgb8888
That seems like overkill and unlikely to actually help anyone, it'd just give userspace more options to shoot itself in the foot with a sub-optimal format choice. And it's a pile of code.
- Hector
Quoted 17 lines · Hector Martin
Q: Why not just add a conversion from XRGB2101010 to XRGB8888?
A: Because that would only fix KDE, and would make it slower vs. not advertising XRGB2101010 at all (double conversions, plus kernel conversion can be slower). Plus, it doesn't make any sense as it only fills in one entry in the conversion matrix. If we wanted to actually fill out the conversion matrix, and thus support everything simpledrm has advertised to day correctly, we would need helpers for:
rgb565->rgb888rgb888->rgb565rgb565->xrgb2101010rgb888->xrgb2101010xrgb2101010->rgb565xrgb2101010->rgb888xrgb2101010->xrgb8888
Quoted 3 lines
That seems like overkill and unlikely to actually help anyone, it'd just give userspace more options to shoot itself in the foot with a sub-optimal format choice. And it's a pile of code.
In addition to everything you mentioned, converting from XRGB2101010 to XRGB8888 loses the additional information, defeating the only point of using XRGB2101010 instead of XRGB8888 in the first place.
Signature · 2 lines
Earthling Michel Dänzer | https://redhat.com Libre software enthusiast | Mesa and Xwayland developer
Thomas Zimmermann
maintainerHi
Am 27.10.22 um 12:13 schrieb Hector Martin:
Quoted 31 lines
Until now, simpledrm unconditionally advertised all formats that can be supported natively as conversions. However, we don't actually have a full conversion matrix of helpers. Although the list is arguably provided to userspace in precedence order, userspace can pick something out-of-order (and thus break when it shouldn't), or simply only support a format that is unsupported (and thus think it can work, which results in the appearance of a hang as FB blits fail later on, instead of the initialization error you'd expect in this case).
Split up the format table into separate ones for each required subset, and then pick one based on the native format. Also remove the native<->conversion overlap check from the helper (which doesn't make sense any more, since the native format is advertised anyway and this way RGB565/RGB888 can share a format table), and instead print the same message in simpledrm when the native format is not one for which we have conversions at all.
This fixes a real user regression where the ?RGB2101010 support commit started advertising it unconditionally where not supported, and KWin decided to start to use it over the native format, but also the fixes the spurious RGB565/RGB888 formats which have been wrongly unconditionally advertised since the dawn of simpledrm.
Note: this patch is merged because splitting it into two patches, one for the helper and one for simpledrm, would regress at the midpoint regardless of the order. If simpledrm is changed first, that would break working conversions to RGB565/RGB888 (since those share a table that does not include the native formats). If the helper is changed first, it would start spuriously advertising all conversion formats when the native format doesn't have any supported conversions at all.
Quoted 8 lines
Acked-by Pekka Paalanen <[email protected]>Fixes 6ea966fca084 ("drm/simpledrm: Add [AX]RGB2101010 formats")Fixes 11e8f5fd223b ("drm: Add simpledrm driver")Signed-off-by Hector Martin <[email protected]>--- drivers/gpu/drm/drm_format_helper.c | 15 ------- drivers/gpu/drm/tiny/simpledrm.c | 62 +++++++++++++++++++++++++----
We currently have two DRM drivers that call drm_fb_build_fourcc_list(): simpledrm and ofdrm. I've been very careful to keep the format selection in sync between them. (That's the reason why the helper exists at all.) If the drivers start to use different logic, it will only become more chaotic.
The format array of ofdrm is at [1]. At a minimum, ofdrm should get the same fix as simpledrm.
[1] https://cgit.freedesktop.org/drm/drm-tip/tree/drivers/gpu/drm/tiny/ofdrm.c#n760
Quoted 103 lines
2 files changed, 55 insertions(+), 22 deletions(-)drivers/gpu/drm/drm_format_helper.c +0
-15@@ -864,20 +864,6 @@ size_t drm_fb_build_fourcc_list(struct drm_device *dev, ++fourccs; } -/* -* The plane's atomic_update helper converts the framebuffer's color format -* to a native format when copying to device memory. -* -* If there is not a single format supported by both, device and -* driver, the native formats are likely not supported by the conversion -* helpers. Therefore *only* support the native formats and add a -* conversion helper ASAP. -*/ -if (!found_native) { -drm_warn (dev,"Format conversion helpers required to add extra formats.\n" ); - goto out; - } -/* * The extra formats, emulated by the driver, go second. */ @@ -898,7 +884,6 @@ size_t drm_fb_build_fourcc_list(struct drm_device *dev, ++fourccs; } -out:return fourccs - fourccs_out; }EXPORT_SYMBOL (drm_fb_build_fourcc_list);drivers/gpu/drm/tiny/simpledrm.c +32
-5@@ -446,22 +446,48 @@ static int simpledrm_device_init_regulators(struct simpledrm_device *sdev)*/ /* -* Support all formats of simplefb and maybe more; in order -* of preference. The display's update function will do any +* Support the subset of formats that we have conversion helpers for, +* in order of preference. The display's update function will do any * conversion necessary. * * TODO: Add blit helpers for remaining formats and uncomment * constants. */ -static const uint32_t simpledrm_primary_plane_formats[] = { + +/* +* Supported conversions to RGB565 and RGB888: +* from [AX]RGB8888 +*/ +static const uint32_t simpledrm_primary_plane_formats_base[] = { +DRM_FORMAT_XRGB8888 , +DRM_FORMAT_ARGB8888 , +}; + +/* +* Supported conversions to [AX]RGB8888: +* A/X variants (no-op) +* from RGB565 +* from RGB888 +*/ +static const uint32_t simpledrm_primary_plane_formats_xrgb8888[] = {DRM_FORMAT_XRGB8888 ,DRM_FORMAT_ARGB8888 , +DRM_FORMAT_RGB888 ,DRM_FORMAT_RGB565 ,//DRM_FORMAT_XRGB1555, //DRM_FORMAT_ARGB1555, -DRM_FORMAT_RGB888 , +}; + +/* +* Supported conversions to [AX]RGB2101010: +* A/X variants (no-op) +* from [AX]RGB8888 +*/ +static const uint32_t simpledrm_primary_plane_formats_xrgb2101010[] = {DRM_FORMAT_XRGB2101010 ,DRM_FORMAT_ARGB2101010 , +DRM_FORMAT_XRGB8888 , +DRM_FORMAT_ARGB8888 , };static const uint64_t simpledrm_primary_plane_format_modifiers[] = {@@ -642,7 +668,8 @@ static struct simpledrm_device *simpledrm_device_create(struct drm_driver *drv,struct drm_encoder *encoder;struct drm_connector *connector;unsigned long max_width, max_height; -size_t nformats; +const uint32_t *conv_formats; +size_t conv_nformats, nformats;int ret; sdev = devm_drm_dev_alloc(&pdev->dev, drv, struct simpledrm_device, dev);
drivers/gpu/drm/tiny/simpledrm.c static struct simpledrm_device *simpledrm_device_create(struct drm_driver *drv, · line 785
Quoted 4 lines
drivers/gpu/drm/tiny/simpledrm.c +1
-0@@ -755,10 +782,31 @@ static struct simpledrm_device *simpledrm_device_create(struct drm_driver *drv, dev->mode_config.funcs = &simpledrm_mode_config_funcs;/* Primary plane */ +switch (format->format) {
I trust you when you say that <native>->XRGB8888 is not enough. But although I've read your replies, I still don't understand why this switch is necessary.
Why don't we call drm_fb_build_fourcc_list() with the native format/formats and let it append a number of formats, such as adding XRGB888, adding ARGB8888 if necessary, adding ARGB2101010 if necessary. Each with a elaborate comment why and which userspace needs the format. (?)
Best regards
Thomas
Quoted 29 lines
drivers/gpu/drm/tiny/simpledrm.c +22
-2+case DRM_FORMAT_RGB565 : +case DRM_FORMAT_RGB888 : + conv_formats = simpledrm_primary_plane_formats_base; + conv_nformats =ARRAY_SIZE (simpledrm_primary_plane_formats_base); +break ; +case DRM_FORMAT_XRGB8888 : +case DRM_FORMAT_ARGB8888 : + conv_formats = simpledrm_primary_plane_formats_xrgb8888; + conv_nformats =ARRAY_SIZE (simpledrm_primary_plane_formats_xrgb8888); +break ; +case DRM_FORMAT_XRGB2101010 : +case DRM_FORMAT_ARGB2101010 : + conv_formats = simpledrm_primary_plane_formats_xrgb2101010; + conv_nformats =ARRAY_SIZE (simpledrm_primary_plane_formats_xrgb2101010); +break ; +default : + conv_formats =NULL ; + conv_nformats =0 ; +drm_warn (dev,"Format conversion helpers required to add extra formats.\n" ); +break ; + } nformats =drm_fb_build_fourcc_list (dev, &format->format,1 , - simpledrm_primary_plane_formats, -ARRAY_SIZE (simpledrm_primary_plane_formats), + conv_formats, conv_nformats, sdev->formats,ARRAY_SIZE (sdev->formats)); primary_plane = &sdev->primary_plane;
Signature · 6 lines
Thomas Zimmermann Graphics Driver Developer SUSE Software Solutions Germany GmbH Maxfeldstr. 5, 90409 Nürnberg, Germany (HRB 36809, AG Nürnberg) Geschäftsführer: Ivo Totev
Quoted 49 lines · On Thu, 27 Oct 2022 13:08:24 +0200 Thomas Zimmermann <[email protected]>…
Hi
Am 27.10.22 um 12:13 schrieb Hector Martin:
Quoted 31 lines
Until now, simpledrm unconditionally advertised all formats that can be supported natively as conversions. However, we don't actually have a full conversion matrix of helpers. Although the list is arguably provided to userspace in precedence order, userspace can pick something out-of-order (and thus break when it shouldn't), or simply only support a format that is unsupported (and thus think it can work, which results in the appearance of a hang as FB blits fail later on, instead of the initialization error you'd expect in this case).
Split up the format table into separate ones for each required subset, and then pick one based on the native format. Also remove the native<->conversion overlap check from the helper (which doesn't make sense any more, since the native format is advertised anyway and this way RGB565/RGB888 can share a format table), and instead print the same message in simpledrm when the native format is not one for which we have conversions at all.
This fixes a real user regression where the ?RGB2101010 support commit started advertising it unconditionally where not supported, and KWin decided to start to use it over the native format, but also the fixes the spurious RGB565/RGB888 formats which have been wrongly unconditionally advertised since the dawn of simpledrm.
Note: this patch is merged because splitting it into two patches, one for the helper and one for simpledrm, would regress at the midpoint regardless of the order. If simpledrm is changed first, that would break working conversions to RGB565/RGB888 (since those share a table that does not include the native formats). If the helper is changed first, it would start spuriously advertising all conversion formats when the native format doesn't have any supported conversions at all.
Quoted 8 lines
Acked-by Pekka Paalanen <[email protected]>Fixes 6ea966fca084 ("drm/simpledrm: Add [AX]RGB2101010 formats")Fixes 11e8f5fd223b ("drm: Add simpledrm driver")Signed-off-by Hector Martin <[email protected]>--- drivers/gpu/drm/drm_format_helper.c | 15 ------- drivers/gpu/drm/tiny/simpledrm.c | 62 +++++++++++++++++++++++++----We currently have two DRM drivers that call
drm_fb_build_fourcc_list():
simpledrm and ofdrm. I've been very careful to keep the format selection in sync between them. (That's the reason why the helper exists at all.) If the drivers start to use different logic, it will only become more chaotic.
Quoted 5 lines
The format array of ofdrm is at [1]. At a minimum, ofdrm should get the same fix as simpledrm.
[1]
https://cgit.freedesktop.org/drm/drm-tip/tree/drivers/gpu/drm/tiny/ofdrm.c#n760
Hi Thomas,
yes, the principle applies to all drivers except VKMS: do not emulate anything in software unless it must be done to prevent kernel regressions on specific hardware.
ofdrm should indeed do the same.
Quoted 112 lines
Quoted 103 lines
2 files changed, 55 insertions(+), 22 deletions(-)drivers/gpu/drm/drm_format_helper.c +0
-15@@ -864,20 +864,6 @@ size_t drm_fb_build_fourcc_list(struct drm_device *dev, ++fourccs; } -/* -* The plane's atomic_update helper converts the framebuffer's color format -* to a native format when copying to device memory. -* -* If there is not a single format supported by both, device and -* driver, the native formats are likely not supported by the conversion -* helpers. Therefore *only* support the native formats and add a -* conversion helper ASAP. -*/ -if (!found_native) { -drm_warn (dev,"Format conversion helpers required to add extra formats.\n" ); - goto out; - } -/* * The extra formats, emulated by the driver, go second. */ @@ -898,7 +884,6 @@ size_t drm_fb_build_fourcc_list(struct drm_device *dev, ++fourccs; } -out:return fourccs - fourccs_out; }EXPORT_SYMBOL (drm_fb_build_fourcc_list);drivers/gpu/drm/tiny/simpledrm.c +32
-5@@ -446,22 +446,48 @@ static int simpledrm_device_init_regulators(struct simpledrm_device *sdev)*/ /* -* Support all formats of simplefb and maybe more; in order -* of preference. The display's update function will do any +* Support the subset of formats that we have conversion helpers for, +* in order of preference. The display's update function will do any * conversion necessary. * * TODO: Add blit helpers for remaining formats and uncomment * constants. */ -static const uint32_t simpledrm_primary_plane_formats[] = { + +/* +* Supported conversions to RGB565 and RGB888: +* from [AX]RGB8888 +*/ +static const uint32_t simpledrm_primary_plane_formats_base[] = { +DRM_FORMAT_XRGB8888 , +DRM_FORMAT_ARGB8888 , +}; + +/* +* Supported conversions to [AX]RGB8888: +* A/X variants (no-op) +* from RGB565 +* from RGB888 +*/ +static const uint32_t simpledrm_primary_plane_formats_xrgb8888[] = {DRM_FORMAT_XRGB8888 ,DRM_FORMAT_ARGB8888 , +DRM_FORMAT_RGB888 ,DRM_FORMAT_RGB565 ,//DRM_FORMAT_XRGB1555, //DRM_FORMAT_ARGB1555, -DRM_FORMAT_RGB888 , +}; + +/* +* Supported conversions to [AX]RGB2101010: +* A/X variants (no-op) +* from [AX]RGB8888 +*/ +static const uint32_t simpledrm_primary_plane_formats_xrgb2101010[] = {DRM_FORMAT_XRGB2101010 ,DRM_FORMAT_ARGB2101010 , +DRM_FORMAT_XRGB8888 , +DRM_FORMAT_ARGB8888 , };static const uint64_t simpledrm_primary_plane_format_modifiers[] = {@@ -642,7 +668,8 @@ static struct simpledrm_device *simpledrm_device_create(struct drm_driver *drv,struct drm_encoder *encoder;struct drm_connector *connector;unsigned long max_width, max_height; -size_t nformats; +const uint32_t *conv_formats; +size_t conv_nformats, nformats;int ret; sdev = devm_drm_dev_alloc(&pdev->dev, drv, struct simpledrm_device, dev);drivers/gpu/drm/tiny/simpledrm.c static struct simpledrm_device *simpledrm_device_create(struct drm_driver *drv, · line 785
Quoted 4 lines
drivers/gpu/drm/tiny/simpledrm.c +1
-0@@ -755,10 +782,31 @@ static struct simpledrm_device *simpledrm_device_create(struct drm_driver *drv, dev->mode_config.funcs = &simpledrm_mode_config_funcs;/* Primary plane */ +switch (format->format) {I trust you when you say that <native>->XRGB8888 is not enough. But although I've read your replies, I still don't understand why this switch is necessary.
Quoted 4 lines
Why don't we call
drm_fb_build_fourcc_list()with the native format/formats and let it append a number of formats, such as adding XRGB888, adding ARGB8888 if necessary, adding ARGB2101010 if necessary. Each with a elaborate comment why and which userspace needs the format. (?)
Something like
uint32_t conv_formats[] = {
DRM_FORMAT_XRGB8888, /* expected by old userspace */
DRM_FORMAT_ARGB8888, /* historically exposed and working */
0,
0,
0,
};size_t conv_nformats = 2;
if (native_format == DRM_FORMAT_XRGB2101010)
conv_formats[conv_nformats++] = DRM_FORMAT_ARGB2101010; /* historically exposed and working */
if (native_format == DRM_FORMAT_XRGB8888) {
conv_formats[conv_nformats++] = DRM_FORMAT_RGB565; /* historically exposed and working */ conv_formats[conv_nformats++] = DRM_FORMAT_RGB888; /* historically exposed and working */
}
maybe?
Thanks,
pq
Quoted 34 lines
Best regards
ThomasQuoted 29 lines
drivers/gpu/drm/tiny/simpledrm.c +22
-2+case DRM_FORMAT_RGB565 : +case DRM_FORMAT_RGB888 : + conv_formats = simpledrm_primary_plane_formats_base; + conv_nformats =ARRAY_SIZE (simpledrm_primary_plane_formats_base); +break ; +case DRM_FORMAT_XRGB8888 : +case DRM_FORMAT_ARGB8888 : + conv_formats = simpledrm_primary_plane_formats_xrgb8888; + conv_nformats =ARRAY_SIZE (simpledrm_primary_plane_formats_xrgb8888); +break ; +case DRM_FORMAT_XRGB2101010 : +case DRM_FORMAT_ARGB2101010 : + conv_formats = simpledrm_primary_plane_formats_xrgb2101010; + conv_nformats =ARRAY_SIZE (simpledrm_primary_plane_formats_xrgb2101010); +break ; +default : + conv_formats =NULL ; + conv_nformats =0 ; +drm_warn (dev,"Format conversion helpers required to add extra formats.\n" ); +break ; + } nformats =drm_fb_build_fourcc_list (dev, &format->format,1 , - simpledrm_primary_plane_formats, -ARRAY_SIZE (simpledrm_primary_plane_formats), + conv_formats, conv_nformats, sdev->formats,ARRAY_SIZE (sdev->formats)); primary_plane = &sdev->primary_plane;
Thomas Zimmermann
maintainerHi
Am 27.10.22 um 14:22 schrieb Pekka Paalanen:
Quoted 197 lines
Quoted 49 lines · On Thu, 27 Oct 2022 13:08:24 +0200 Thomas Zimmermann <[email protected]>…
Hi
Am 27.10.22 um 12:13 schrieb Hector Martin:
Quoted 31 lines
Until now, simpledrm unconditionally advertised all formats that can be supported natively as conversions. However, we don't actually have a full conversion matrix of helpers. Although the list is arguably provided to userspace in precedence order, userspace can pick something out-of-order (and thus break when it shouldn't), or simply only support a format that is unsupported (and thus think it can work, which results in the appearance of a hang as FB blits fail later on, instead of the initialization error you'd expect in this case).
Split up the format table into separate ones for each required subset, and then pick one based on the native format. Also remove the native<->conversion overlap check from the helper (which doesn't make sense any more, since the native format is advertised anyway and this way RGB565/RGB888 can share a format table), and instead print the same message in simpledrm when the native format is not one for which we have conversions at all.
This fixes a real user regression where the ?RGB2101010 support commit started advertising it unconditionally where not supported, and KWin decided to start to use it over the native format, but also the fixes the spurious RGB565/RGB888 formats which have been wrongly unconditionally advertised since the dawn of simpledrm.
Note: this patch is merged because splitting it into two patches, one for the helper and one for simpledrm, would regress at the midpoint regardless of the order. If simpledrm is changed first, that would break working conversions to RGB565/RGB888 (since those share a table that does not include the native formats). If the helper is changed first, it would start spuriously advertising all conversion formats when the native format doesn't have any supported conversions at all.
Quoted 8 lines
Acked-by Pekka Paalanen <[email protected]>Fixes 6ea966fca084 ("drm/simpledrm: Add [AX]RGB2101010 formats")Fixes 11e8f5fd223b ("drm: Add simpledrm driver")Signed-off-by Hector Martin <[email protected]>--- drivers/gpu/drm/drm_format_helper.c | 15 ------- drivers/gpu/drm/tiny/simpledrm.c | 62 +++++++++++++++++++++++++----We currently have two DRM drivers that call
drm_fb_build_fourcc_list():
simpledrm and ofdrm. I've been very careful to keep the format selection in sync between them. (That's the reason why the helper exists at all.) If the drivers start to use different logic, it will only become more chaotic.Quoted 5 lines
The format array of ofdrm is at [1]. At a minimum, ofdrm should get the same fix as simpledrm.
[1]
https://cgit.freedesktop.org/drm/drm-tip/tree/drivers/gpu/drm/tiny/ofdrm.c#n760Hi Thomas,
yes, the principle applies to all drivers except VKMS: do not emulate anything in software unless it must be done to prevent kernel regressions on specific hardware.
ofdrm should indeed do the same.
Quoted 112 lines
Quoted 103 lines
2 files changed, 55 insertions(+), 22 deletions(-)drivers/gpu/drm/drm_format_helper.c +0
-15@@ -864,20 +864,6 @@ size_t drm_fb_build_fourcc_list(struct drm_device *dev, ++fourccs; } -/* -* The plane's atomic_update helper converts the framebuffer's color format -* to a native format when copying to device memory. -* -* If there is not a single format supported by both, device and -* driver, the native formats are likely not supported by the conversion -* helpers. Therefore *only* support the native formats and add a -* conversion helper ASAP. -*/ -if (!found_native) { -drm_warn (dev,"Format conversion helpers required to add extra formats.\n" ); - goto out; - } -/* * The extra formats, emulated by the driver, go second. */ @@ -898,7 +884,6 @@ size_t drm_fb_build_fourcc_list(struct drm_device *dev, ++fourccs; } -out:return fourccs - fourccs_out; }EXPORT_SYMBOL (drm_fb_build_fourcc_list);drivers/gpu/drm/tiny/simpledrm.c +32
-5@@ -446,22 +446,48 @@ static int simpledrm_device_init_regulators(struct simpledrm_device *sdev)*/ /* -* Support all formats of simplefb and maybe more; in order -* of preference. The display's update function will do any +* Support the subset of formats that we have conversion helpers for, +* in order of preference. The display's update function will do any * conversion necessary. * * TODO: Add blit helpers for remaining formats and uncomment * constants. */ -static const uint32_t simpledrm_primary_plane_formats[] = { + +/* +* Supported conversions to RGB565 and RGB888: +* from [AX]RGB8888 +*/ +static const uint32_t simpledrm_primary_plane_formats_base[] = { +DRM_FORMAT_XRGB8888 , +DRM_FORMAT_ARGB8888 , +}; + +/* +* Supported conversions to [AX]RGB8888: +* A/X variants (no-op) +* from RGB565 +* from RGB888 +*/ +static const uint32_t simpledrm_primary_plane_formats_xrgb8888[] = {DRM_FORMAT_XRGB8888 ,DRM_FORMAT_ARGB8888 , +DRM_FORMAT_RGB888 ,DRM_FORMAT_RGB565 ,//DRM_FORMAT_XRGB1555, //DRM_FORMAT_ARGB1555, -DRM_FORMAT_RGB888 , +}; + +/* +* Supported conversions to [AX]RGB2101010: +* A/X variants (no-op) +* from [AX]RGB8888 +*/ +static const uint32_t simpledrm_primary_plane_formats_xrgb2101010[] = {DRM_FORMAT_XRGB2101010 ,DRM_FORMAT_ARGB2101010 , +DRM_FORMAT_XRGB8888 , +DRM_FORMAT_ARGB8888 , };static const uint64_t simpledrm_primary_plane_format_modifiers[] = {@@ -642,7 +668,8 @@ static struct simpledrm_device *simpledrm_device_create(struct drm_driver *drv,struct drm_encoder *encoder;struct drm_connector *connector;unsigned long max_width, max_height; -size_t nformats; +const uint32_t *conv_formats; +size_t conv_nformats, nformats;int ret; sdev = devm_drm_dev_alloc(&pdev->dev, drv, struct simpledrm_device, dev);drivers/gpu/drm/tiny/simpledrm.c static struct simpledrm_device *simpledrm_device_create(struct drm_driver *drv, · line 785
Quoted 4 lines
drivers/gpu/drm/tiny/simpledrm.c +1
-0@@ -755,10 +782,31 @@ static struct simpledrm_device *simpledrm_device_create(struct drm_driver *drv, dev->mode_config.funcs = &simpledrm_mode_config_funcs;/* Primary plane */ +switch (format->format) {I trust you when you say that <native>->XRGB8888 is not enough. But although I've read your replies, I still don't understand why this switch is necessary.
Quoted 4 lines
Why don't we call
drm_fb_build_fourcc_list()with the native format/formats and let it append a number of formats, such as adding XRGB888, adding ARGB8888 if necessary, adding ARGB2101010 if necessary. Each with a elaborate comment why and which userspace needs the format. (?)Something like
uint32_t conv_formats[] = { DRM_FORMAT_XRGB8888, /* expected by old userspace */ DRM_FORMAT_ARGB8888, /* historically exposed and working */ 0, 0, 0, };
size_tconv_nformats= 2;if (
native_format==DRM_FORMAT_XRGB2101010)
conv_formats[conv_nformats++] =DRM_FORMAT_ARGB2101010; /* historically exposed and working */
Quoted 6 lines
if (
native_format==DRM_FORMAT_XRGB8888) {
conv_formats[conv_nformats++] =DRM_FORMAT_RGB565; /* historically exposed and working */ conv_formats[conv_nformats++] =DRM_FORMAT_RGB888; /* historically exposed and working */
}maybe?
Yes, that's what I have in mind. We give a list of native formats to drm_fb_build_fourcc_list(), the helper appends whatever is needed and returns the result for use by the driver.
Currently, drm_fb_build_fourcc_list() gets the native formats plus a list of all driver-exported formats. And if I'm not mistake, we could remove the driver list entirely.
Maybe we could condense the native-format list to a single entry. This would simplify things significantly.
Best regards
Thomas
Quoted 42 lines
Thanks,
pqQuoted 34 lines
Best regards
ThomasQuoted 29 lines
drivers/gpu/drm/tiny/simpledrm.c +22
-2+case DRM_FORMAT_RGB565 : +case DRM_FORMAT_RGB888 : + conv_formats = simpledrm_primary_plane_formats_base; + conv_nformats =ARRAY_SIZE (simpledrm_primary_plane_formats_base); +break ; +case DRM_FORMAT_XRGB8888 : +case DRM_FORMAT_ARGB8888 : + conv_formats = simpledrm_primary_plane_formats_xrgb8888; + conv_nformats =ARRAY_SIZE (simpledrm_primary_plane_formats_xrgb8888); +break ; +case DRM_FORMAT_XRGB2101010 : +case DRM_FORMAT_ARGB2101010 : + conv_formats = simpledrm_primary_plane_formats_xrgb2101010; + conv_nformats =ARRAY_SIZE (simpledrm_primary_plane_formats_xrgb2101010); +break ; +default : + conv_formats =NULL ; + conv_nformats =0 ; +drm_warn (dev,"Format conversion helpers required to add extra formats.\n" ); +break ; + } nformats =drm_fb_build_fourcc_list (dev, &format->format,1 , - simpledrm_primary_plane_formats, -ARRAY_SIZE (simpledrm_primary_plane_formats), + conv_formats, conv_nformats, sdev->formats,ARRAY_SIZE (sdev->formats)); primary_plane = &sdev->primary_plane;
Signature · 6 lines
Thomas Zimmermann Graphics Driver Developer SUSE Software Solutions Germany GmbH Maxfeldstr. 5, 90409 Nürnberg, Germany (HRB 36809, AG Nürnberg) Geschäftsführer: Ivo Totev
Quoted 8 lines · Thomas Zimmermann
We currently have two DRM drivers that call
drm_fb_build_fourcc_list():
simpledrm and ofdrm. I've been very careful to keep the format selection in sync between them. (That's the reason why the helper exists at all.) If the drivers start to use different logic, it will only become more chaotic.The format array of ofdrm is at [1]. At a minimum, ofdrm should get the same fix as simpledrm.
Looks like this was merged recently, so I didn't see it on my tree (I was basing off of 6.1-rc2).
Since this patch is a regression fix, it should be applied to drm-fixes (and automatically picked up by stable folks) soon to be fixed in 6.1, and then we can fix whatever is needed in ofdrm separately in drm-tip. As long as ofdrm is ready for the new behavior prior to the merge of drm-tip with 6.1, there will be no breakage.
In this case, the change required to ofdrm is probably just to replace that array with just DRM_FORMAT_XRGB8888 (which should be the only supported fallback format for new drivers) and then to add a test to only expose it for formats for which we actually have conversion helpers, similar to what the the switch() enumerates here. That logic could later be moved into the helper as a refactor.
Quoted 11 lines
drivers/gpu/drm/tiny/simpledrm.c static struct simpledrm_device *simpledrm_device_create(struct drm_driver *drv, · line 785
Quoted 2 lines
/* Primary plane */
drivers/gpu/drm/tiny/simpledrm.c +1
-0+switch (format->format) {I trust you when you say that <native>->XRGB8888 is not enough. But although I've read your replies, I still don't understand why this switch is necessary.
Why don't we call
drm_fb_build_fourcc_list()with the native format/formats and let it append a number of formats, such as adding XRGB888, adding ARGB8888 if necessary, adding ARGB2101010 if necessary. Each with a elaborate comment why and which userspace needs the format. (?)
That would be fine to do, it would just be moving the logic to the helper. That kind of refactoring is better suited for subsequent patches. This is a regression fix, it attempts to minimize the amount of refactoring, which means keeping the logic in simpledrm, to make it easier to review for correctness.
Also, that would change the API of that function, which would likely make the merge with the new ofdrm painful. The way things are now, a small fix to ofdrm will make it compatible with both the state before and after this patch, which means the merge will go through painlessly, and then we can just refactor everything once everything is in the same tree.
- Hector
Quoted 8 lines · Thomas Zimmermann
I trust you when you say that <native>->XRGB8888 is not enough. But although I've read your replies, I still don't understand why this switch is necessary.
Why don't we call
drm_fb_build_fourcc_list()with the native format/formats and let it append a number of formats, such as adding XRGB888, adding ARGB8888 if necessary, adding ARGB2101010 if necessary. Each with a elaborate comment why and which userspace needs the format. (?)
Are you saying there is some real userspace that breaks without the alpha formats? That would already be broken on many devices.
Signature · 2 lines
Ville Syrjälä Intel
Thomas Zimmermann
maintainerHi
Am 28.10.22 um 12:04 schrieb Ville Syrjälä:
Quoted 5 lines · Thomas Zimmermann
Quoted 4 lines
I trust you when you say that <native>->XRGB8888 is not enough. But although I've read your replies, I still don't understand why this switch is necessary.
Quoted 7 lines
Quoted 4 lines
Why don't we call
drm_fb_build_fourcc_list()with the native format/formats and let it append a number of formats, such as adding XRGB888, adding ARGB8888 if necessary, adding ARGB2101010 if necessary. Each with a elaborate comment why and which userspace needs the format. (?)Are you saying there is some real userspace that breaks without the alpha formats? That would already be broken on many devices.
We should scrap them all IMHO, unless the hardware really supports them. See the other discussion on this patch's v2 as well.
Best regards
Thomas
Quoted 1 line
Signature · 6 lines
Thomas Zimmermann Graphics Driver Developer SUSE Software Solutions Germany GmbH Maxfeldstr. 5, 90409 Nürnberg, Germany (HRB 36809, AG Nürnberg) Geschäftsführer: Ivo Totev