[Buildroot] [PATCH 1/1] ti-sgx-*: bump from SDK 02.00 to SDK 04.00
tolvupostur at gmail.com
Mon May 28 12:20:44 UTC 2018
I’m on holidays until Wednesday with only mobile devices, so if someone
else wants to step in, please do.
On Sunday, 27 May 2018, Romain Naour <romain.naour at gmail.com> wrote:
> Hi Einar,
> Le 24/05/2018 à 17:38, Einar Jón a écrit :
> > Hi,
> > Just to explain this one. The ti-sgx-* drivers are very old, but
> > nothing is using them except beaglebone_qt5_defconfig.
> > I bumped beaglebone_qt5_defconfig to the same level as
> > beaglebone_defconfig (the SDK 4.00 version in
> > http://www.ti.com/tool/PROCESSOR-SDK-AM335X)
> > The ti-sdx-* drivers have been bumped to the same SDK 4.00 version.
> > The ti-sdx-um has a dependency on a new package ti-sdx-libgdm, which
> > in turn depends on wayland.
> > This again gives QT a valid EGL backend for beaglebone_qt5.
> > Let me know if this should be done differently.
> Thanks for this patch. We will need it soon in order to remove the gcc 5
> dependency in beaglebone_qt5 defconfig.
> I'm not sure if we must bump all packages related to the TI SDK or bump all
> packages one by one to ease the review.
I did this in a single commit since any split (except a standalone
ti-sgx-libgbm) would be a commit with a broken build. The places I’ve
worked for for the last 5 years hate those.
> This patch add a new ti-sgx-libgbm package, it should be added in a
> separate patch.
> Also, BR2_PACKAGE_TI_SGX_KM_AM4430 and BR2_PACKAGE_TI_SGX_KM_AM5430
> symbols are
> removed. It must be handled in Config.in.legacy.
True. I didn’t think of that since none of the defconfigs use it. I can see
that now there is a legacy patch for the _KM addition.
Also, the new _KM addition to these keywords makes these parameters a bit
weird, because it is used in both ti-sgx-km and ti-sgx-um.
Maybe 3 _UM bools should be made that are just set to the same value as the
> Why S80ti-sgx is removed ?
Because it is both unneeded and obsolete.
The current repo of ti-sgx-um has 3 subfolders, one for each
BR2_PACKAGE_TI_SGX_KM_AMxxx option. Each one has a sysV init file (called
rc-???) that is very different from the one I deleted.
Now I simply rename rc-??? to S80ti-sgx in the post-hook that previously
copied it from package/ti-sgx-um.
It actually is a very ugly $(INSTALL) && $(RM), since it seems weird to
“mv” files on target/. And I copy it from staging instead of target to be
sure that it always exists - but now I realize that I did that because I
was running the hook twice (as post_install and sysv_init).
> > Best regards,
> > Einar Jón
> > On 24 May 2018 at 17:24, Einar Jon Gunnarsson <tolvupostur at gmail.com>
> >> Added new package ti-sgx-libgbm
> >> Fixed build and added license file hashes.
> >> Bumped beaglebone_qt5 to same level as beaglebone
> >> Signed-off-by: Einar Jon Gunnarsson <tolvupostur at gmail.com>
> >> ---
> >> ...nux-4.1-sgx.fragment => linux-4.9-sgx.fragment} | 0
> >> configs/beaglebone_qt5_defconfig | 14 +++++------
> >> package/Config.in | 1 +
> >> package/ti-sgx-demos/ti-sgx-demos.hash | 3 ++-
> >> package/ti-sgx-demos/ti-sgx-demos.mk | 20
> >> package/ti-sgx-km/Config.in | 17 +++++--------
> >> package/ti-sgx-km/ti-sgx-km.hash | 3 ++-
> >> package/ti-sgx-km/ti-sgx-km.mk | 22
> >> package/ti-sgx-libgbm/Config.in | 9 +++++++
> >> package/ti-sgx-libgbm/ti-sgx-libgbm.hash | 3 +++
> >> package/ti-sgx-libgbm/ti-sgx-libgbm.mk | 17 +++++++++++++
> >> package/ti-sgx-um/Config.in | 3 ++-
> >> package/ti-sgx-um/S80ti-sgx | 14 -----------
> >> package/ti-sgx-um/powervr.ini | 4 +--
> >> package/ti-sgx-um/ti-sgx-um.hash | 3 ++-
> >> package/ti-sgx-um/ti-sgx-um.mk | 29
+31 610 957234
-------------- next part --------------
An HTML attachment was scrubbed...
More information about the buildroot