Message ID | 20230427210746.1915362-1-rs@ti.com |
---|---|
Headers | show |
Series | Enable vulkan proper | expand |
On Thu, Apr 27, 2023 at 04:07:42PM -0500, rs@ti.com wrote: > From: Randolph Sapp <rs@ti.com> > > Properly enable Vulkan with all WSI extensions enabled. Attempt some > light packaging changes to the graphics umlibs recipe so we can decrease > the number of check we have to skip, and the checks become a little more > targeted. > > This version uses a cairo append that conditionally tweaks PACKAGECONFIG > flags based on the GPU driver that is selected. We're using this driver > flag instead of a machine feature because the open source driver will > add a whole new set of preferred configs that will be optimized later. LGTM. Is it kirkstone-specific? Any plans for master? > Randolph Sapp (4): > ti-img-rogue-umlibs: bump srcrev, add vulkan, update deps > pvr-mesa: add new vulkan configuration options and wsi > cairo: resolve opengl version > ti-img-rogue-umlibs: package firmware separately > > .../recipes-graphics/cairo/cairo_%.bbappend | 1 + > .../recipes-graphics/cairo/pvr-cairo.inc | 5 +++ > .../ti-img-rogue-umlibs_23.1.6404501.bb | 33 +++++++++++++++---- > .../recipes-graphics/mesa/pvr-mesa.inc | 6 +++- > 4 files changed, 37 insertions(+), 8 deletions(-) > create mode 100644 meta-ti-bsp/recipes-graphics/cairo/cairo_%.bbappend > create mode 100644 meta-ti-bsp/recipes-graphics/cairo/pvr-cairo.inc
On 4/27/23 16:50, Denys Dmytriyenko wrote: > On Thu, Apr 27, 2023 at 04:07:42PM -0500, rs@ti.com wrote: >> From: Randolph Sapp <rs@ti.com> >> >> Properly enable Vulkan with all WSI extensions enabled. Attempt some >> light packaging changes to the graphics umlibs recipe so we can decrease >> the number of check we have to skip, and the checks become a little more >> targeted. >> >> This version uses a cairo append that conditionally tweaks PACKAGECONFIG >> flags based on the GPU driver that is selected. We're using this driver >> flag instead of a machine feature because the open source driver will >> add a whole new set of preferred configs that will be optimized later. > > LGTM. Is it kirkstone-specific? Any plans for master? Yeah, unfortunately this is kirkstone specific for now. Master version soon. Just wanted to get this and the meta-arago patches sorted and merged first. > >> Randolph Sapp (4): >> ti-img-rogue-umlibs: bump srcrev, add vulkan, update deps >> pvr-mesa: add new vulkan configuration options and wsi >> cairo: resolve opengl version >> ti-img-rogue-umlibs: package firmware separately >> >> .../recipes-graphics/cairo/cairo_%.bbappend | 1 + >> .../recipes-graphics/cairo/pvr-cairo.inc | 5 +++ >> .../ti-img-rogue-umlibs_23.1.6404501.bb | 33 +++++++++++++++---- >> .../recipes-graphics/mesa/pvr-mesa.inc | 6 +++- >> 4 files changed, 37 insertions(+), 8 deletions(-) >> create mode 100644 meta-ti-bsp/recipes-graphics/cairo/cairo_%.bbappend >> create mode 100644 meta-ti-bsp/recipes-graphics/cairo/pvr-cairo.inc
On 4/27/23 16:50, Denys Dmytriyenko wrote: > On Thu, Apr 27, 2023 at 04:07:42PM -0500, rs@ti.com wrote: >> From: Randolph Sapp <rs@ti.com> >> >> Properly enable Vulkan with all WSI extensions enabled. Attempt some >> light packaging changes to the graphics umlibs recipe so we can decrease >> the number of check we have to skip, and the checks become a little more >> targeted. >> >> This version uses a cairo append that conditionally tweaks PACKAGECONFIG >> flags based on the GPU driver that is selected. We're using this driver >> flag instead of a machine feature because the open source driver will >> add a whole new set of preferred configs that will be optimized later. > > LGTM. Is it kirkstone-specific? Any plans for master? Actually, disregard that last response. It'll apply cleanly to master if you also cherry-pick the following patches from kirkstone: c994d48021664d328ced52829542a3b1b86c4e02 2cc7895ebfea1df4afc2229f9762f129946b0b06 b86fb13fb9638635bf957156833bccc10ddf5393 I must have incorrectly assumed those wouldn't apply to master when I submitted them originally. My bad, they apply fine. > >> Randolph Sapp (4): >> ti-img-rogue-umlibs: bump srcrev, add vulkan, update deps >> pvr-mesa: add new vulkan configuration options and wsi >> cairo: resolve opengl version >> ti-img-rogue-umlibs: package firmware separately >> >> .../recipes-graphics/cairo/cairo_%.bbappend | 1 + >> .../recipes-graphics/cairo/pvr-cairo.inc | 5 +++ >> .../ti-img-rogue-umlibs_23.1.6404501.bb | 33 +++++++++++++++---- >> .../recipes-graphics/mesa/pvr-mesa.inc | 6 +++- >> 4 files changed, 37 insertions(+), 8 deletions(-) >> create mode 100644 meta-ti-bsp/recipes-graphics/cairo/cairo_%.bbappend >> create mode 100644 meta-ti-bsp/recipes-graphics/cairo/pvr-cairo.inc
On Thu, Apr 27, 2023 at 06:58:32PM -0500, Randolph Sapp via lists.yoctoproject.org wrote: > On 4/27/23 16:50, Denys Dmytriyenko wrote: > >On Thu, Apr 27, 2023 at 04:07:42PM -0500, rs@ti.com wrote: > >>From: Randolph Sapp <rs@ti.com> > >> > >>Properly enable Vulkan with all WSI extensions enabled. Attempt some > >>light packaging changes to the graphics umlibs recipe so we can decrease > >>the number of check we have to skip, and the checks become a little more > >>targeted. > >> > >>This version uses a cairo append that conditionally tweaks PACKAGECONFIG > >>flags based on the GPU driver that is selected. We're using this driver > >>flag instead of a machine feature because the open source driver will > >>add a whole new set of preferred configs that will be optimized later. > > > >LGTM. Is it kirkstone-specific? Any plans for master? > > Actually, disregard that last response. It'll apply cleanly to > master if you also cherry-pick the following patches from kirkstone: > c994d48021664d328ced52829542a3b1b86c4e02 > 2cc7895ebfea1df4afc2229f9762f129946b0b06 > b86fb13fb9638635bf957156833bccc10ddf5393 Can you please be more specific - none of these 3 commits exist in meta-ti nor meta-arago... > I must have incorrectly assumed those wouldn't apply to master when > I submitted them originally. My bad, they apply fine. > > > > >>Randolph Sapp (4): > >> ti-img-rogue-umlibs: bump srcrev, add vulkan, update deps > >> pvr-mesa: add new vulkan configuration options and wsi > >> cairo: resolve opengl version > >> ti-img-rogue-umlibs: package firmware separately > >> > >> .../recipes-graphics/cairo/cairo_%.bbappend | 1 + > >> .../recipes-graphics/cairo/pvr-cairo.inc | 5 +++ > >> .../ti-img-rogue-umlibs_23.1.6404501.bb | 33 +++++++++++++++---- > >> .../recipes-graphics/mesa/pvr-mesa.inc | 6 +++- > >> 4 files changed, 37 insertions(+), 8 deletions(-) > >> create mode 100644 meta-ti-bsp/recipes-graphics/cairo/cairo_%.bbappend > >> create mode 100644 meta-ti-bsp/recipes-graphics/cairo/pvr-cairo.inc
On 4/27/23 20:42, Denys Dmytriyenko wrote: > On Thu, Apr 27, 2023 at 06:58:32PM -0500, Randolph Sapp via lists.yoctoproject.org wrote: >> On 4/27/23 16:50, Denys Dmytriyenko wrote: >>> On Thu, Apr 27, 2023 at 04:07:42PM -0500, rs@ti.com wrote: >>>> From: Randolph Sapp <rs@ti.com> >>>> >>>> Properly enable Vulkan with all WSI extensions enabled. Attempt some >>>> light packaging changes to the graphics umlibs recipe so we can decrease >>>> the number of check we have to skip, and the checks become a little more >>>> targeted. >>>> >>>> This version uses a cairo append that conditionally tweaks PACKAGECONFIG >>>> flags based on the GPU driver that is selected. We're using this driver >>>> flag instead of a machine feature because the open source driver will >>>> add a whole new set of preferred configs that will be optimized later. >>> >>> LGTM. Is it kirkstone-specific? Any plans for master? >> >> Actually, disregard that last response. It'll apply cleanly to >> master if you also cherry-pick the following patches from kirkstone: >> c994d48021664d328ced52829542a3b1b86c4e02 >> 2cc7895ebfea1df4afc2229f9762f129946b0b06 >> b86fb13fb9638635bf957156833bccc10ddf5393 > > Can you please be more specific - none of these 3 commits exist in meta-ti nor > meta-arago... > Right, my bad. I meant this patch set: https://lists.yoctoproject.org/g/meta-ti/message/16397 Or these commits from kirkstone: 16a8078a43eef297449a04d6861fd9811bfeb052 b9581a704648cb3ed70c6bcb28d5a1a89b08534d c8ba8aaade90101f06164aec1b393d7c8b3aa201 I don't know where I got those previous hashes. Must have done something goofy. > >> I must have incorrectly assumed those wouldn't apply to master when >> I submitted them originally. My bad, they apply fine. >> >>> >>>> Randolph Sapp (4): >>>> ti-img-rogue-umlibs: bump srcrev, add vulkan, update deps >>>> pvr-mesa: add new vulkan configuration options and wsi >>>> cairo: resolve opengl version >>>> ti-img-rogue-umlibs: package firmware separately >>>> >>>> .../recipes-graphics/cairo/cairo_%.bbappend | 1 + >>>> .../recipes-graphics/cairo/pvr-cairo.inc | 5 +++ >>>> .../ti-img-rogue-umlibs_23.1.6404501.bb | 33 +++++++++++++++---- >>>> .../recipes-graphics/mesa/pvr-mesa.inc | 6 +++- >>>> 4 files changed, 37 insertions(+), 8 deletions(-) >>>> create mode 100644 meta-ti-bsp/recipes-graphics/cairo/cairo_%.bbappend >>>> create mode 100644 meta-ti-bsp/recipes-graphics/cairo/pvr-cairo.inc
On Fri, Apr 28, 2023 at 11:30:06AM -0500, Randolph Sapp wrote: > On 4/27/23 20:42, Denys Dmytriyenko wrote: > >On Thu, Apr 27, 2023 at 06:58:32PM -0500, Randolph Sapp via lists.yoctoproject.org wrote: > >>On 4/27/23 16:50, Denys Dmytriyenko wrote: > >>>On Thu, Apr 27, 2023 at 04:07:42PM -0500, rs@ti.com wrote: > >>>>From: Randolph Sapp <rs@ti.com> > >>>> > >>>>Properly enable Vulkan with all WSI extensions enabled. Attempt some > >>>>light packaging changes to the graphics umlibs recipe so we can decrease > >>>>the number of check we have to skip, and the checks become a little more > >>>>targeted. > >>>> > >>>>This version uses a cairo append that conditionally tweaks PACKAGECONFIG > >>>>flags based on the GPU driver that is selected. We're using this driver > >>>>flag instead of a machine feature because the open source driver will > >>>>add a whole new set of preferred configs that will be optimized later. > >>> > >>>LGTM. Is it kirkstone-specific? Any plans for master? > >> > >>Actually, disregard that last response. It'll apply cleanly to > >>master if you also cherry-pick the following patches from kirkstone: > >>c994d48021664d328ced52829542a3b1b86c4e02 > >>2cc7895ebfea1df4afc2229f9762f129946b0b06 > >>b86fb13fb9638635bf957156833bccc10ddf5393 > > > >Can you please be more specific - none of these 3 commits exist in meta-ti nor > >meta-arago... > > > > Right, my bad. I meant this patch set: > https://lists.yoctoproject.org/g/meta-ti/message/16397 > > Or these commits from kirkstone: > 16a8078a43eef297449a04d6861fd9811bfeb052 > b9581a704648cb3ed70c6bcb28d5a1a89b08534d > c8ba8aaade90101f06164aec1b393d7c8b3aa201 Thanks. The original patches for these commits were marked specifically for kirkstone (same as this series) and I guess the assumption was that they won't apply to master and were skipped. I just cherry-picked them for master and can confirm everything applies and builds cleanly - I just re-sent them to the list just in case. Once they are in, this Vulkan series can also be applied to master. > I don't know where I got those previous hashes. Must have done > something goofy. > > > > > >>I must have incorrectly assumed those wouldn't apply to master when > >>I submitted them originally. My bad, they apply fine. > >> > >>> > >>>>Randolph Sapp (4): > >>>> ti-img-rogue-umlibs: bump srcrev, add vulkan, update deps > >>>> pvr-mesa: add new vulkan configuration options and wsi > >>>> cairo: resolve opengl version > >>>> ti-img-rogue-umlibs: package firmware separately > >>>> > >>>> .../recipes-graphics/cairo/cairo_%.bbappend | 1 + > >>>> .../recipes-graphics/cairo/pvr-cairo.inc | 5 +++ > >>>> .../ti-img-rogue-umlibs_23.1.6404501.bb | 33 +++++++++++++++---- > >>>> .../recipes-graphics/mesa/pvr-mesa.inc | 6 +++- > >>>> 4 files changed, 37 insertions(+), 8 deletions(-) > >>>> create mode 100644 meta-ti-bsp/recipes-graphics/cairo/cairo_%.bbappend
On 5/1/23 22:46, Denys Dmytriyenko wrote: > On Fri, Apr 28, 2023 at 11:30:06AM -0500, Randolph Sapp wrote: >> On 4/27/23 20:42, Denys Dmytriyenko wrote: >>> On Thu, Apr 27, 2023 at 06:58:32PM -0500, Randolph Sapp via lists.yoctoproject.org wrote: >>>> On 4/27/23 16:50, Denys Dmytriyenko wrote: >>>>> On Thu, Apr 27, 2023 at 04:07:42PM -0500, rs@ti.com wrote: >>>>>> From: Randolph Sapp <rs@ti.com> >>>>>> >>>>>> Properly enable Vulkan with all WSI extensions enabled. Attempt some >>>>>> light packaging changes to the graphics umlibs recipe so we can decrease >>>>>> the number of check we have to skip, and the checks become a little more >>>>>> targeted. >>>>>> >>>>>> This version uses a cairo append that conditionally tweaks PACKAGECONFIG >>>>>> flags based on the GPU driver that is selected. We're using this driver >>>>>> flag instead of a machine feature because the open source driver will >>>>>> add a whole new set of preferred configs that will be optimized later. >>>>> >>>>> LGTM. Is it kirkstone-specific? Any plans for master? >>>> >>>> Actually, disregard that last response. It'll apply cleanly to >>>> master if you also cherry-pick the following patches from kirkstone: >>>> c994d48021664d328ced52829542a3b1b86c4e02 >>>> 2cc7895ebfea1df4afc2229f9762f129946b0b06 >>>> b86fb13fb9638635bf957156833bccc10ddf5393 >>> >>> Can you please be more specific - none of these 3 commits exist in meta-ti nor >>> meta-arago... >>> >> >> Right, my bad. I meant this patch set: >> https://lists.yoctoproject.org/g/meta-ti/message/16397 >> >> Or these commits from kirkstone: >> 16a8078a43eef297449a04d6861fd9811bfeb052 >> b9581a704648cb3ed70c6bcb28d5a1a89b08534d >> c8ba8aaade90101f06164aec1b393d7c8b3aa201 > > Thanks. The original patches for these commits were marked specifically for > kirkstone (same as this series) and I guess the assumption was that they won't > apply to master and were skipped. > > I just cherry-picked them for master and can confirm everything applies and > builds cleanly - I just re-sent them to the list just in case. Once they are > in, this Vulkan series can also be applied to master. > Excellent, thanks Denys! > >> I don't know where I got those previous hashes. Must have done >> something goofy. >> >> >>> >>>> I must have incorrectly assumed those wouldn't apply to master when >>>> I submitted them originally. My bad, they apply fine. >>>> >>>>> >>>>>> Randolph Sapp (4): >>>>>> ti-img-rogue-umlibs: bump srcrev, add vulkan, update deps >>>>>> pvr-mesa: add new vulkan configuration options and wsi >>>>>> cairo: resolve opengl version >>>>>> ti-img-rogue-umlibs: package firmware separately >>>>>> >>>>>> .../recipes-graphics/cairo/cairo_%.bbappend | 1 + >>>>>> .../recipes-graphics/cairo/pvr-cairo.inc | 5 +++ >>>>>> .../ti-img-rogue-umlibs_23.1.6404501.bb | 33 +++++++++++++++---- >>>>>> .../recipes-graphics/mesa/pvr-mesa.inc | 6 +++- >>>>>> 4 files changed, 37 insertions(+), 8 deletions(-) >>>>>> create mode 100644 meta-ti-bsp/recipes-graphics/cairo/cairo_%.bbappend
From: Randolph Sapp <rs@ti.com> Properly enable Vulkan with all WSI extensions enabled. Attempt some light packaging changes to the graphics umlibs recipe so we can decrease the number of check we have to skip, and the checks become a little more targeted. This version uses a cairo append that conditionally tweaks PACKAGECONFIG flags based on the GPU driver that is selected. We're using this driver flag instead of a machine feature because the open source driver will add a whole new set of preferred configs that will be optimized later. Randolph Sapp (4): ti-img-rogue-umlibs: bump srcrev, add vulkan, update deps pvr-mesa: add new vulkan configuration options and wsi cairo: resolve opengl version ti-img-rogue-umlibs: package firmware separately .../recipes-graphics/cairo/cairo_%.bbappend | 1 + .../recipes-graphics/cairo/pvr-cairo.inc | 5 +++ .../ti-img-rogue-umlibs_23.1.6404501.bb | 33 +++++++++++++++---- .../recipes-graphics/mesa/pvr-mesa.inc | 6 +++- 4 files changed, 37 insertions(+), 8 deletions(-) create mode 100644 meta-ti-bsp/recipes-graphics/cairo/cairo_%.bbappend create mode 100644 meta-ti-bsp/recipes-graphics/cairo/pvr-cairo.inc