[meta-ti] [PATCH] u-boot: Alawys build with disabled PIE

[Khem Raj]

Hi Tom

I think, it does for now. Eventually, it might be interesting to see if
u-boot could be compiled as static PIE, looking  at the code generated
by gcc for arm/beaglebone, there is additional penalty since arm gcc
is reloading pc-relative addresses, putting an extra slot in the
constant pool each time which can be significant if we use small
functions as seems to be the case in u-boot.

Thanks
-KhemOn Thu, Oct 18, 2018 at 7:09 AM Tom Rini <trini at konsulko.com> wrote:
>
> On Wed, Oct 17, 2018 at 10:49:46PM -0700, Khem Raj wrote:
> > some OE distros can default to hardened toolchains which means PIE is
> > added implicitly to compiler cmdline, here this options makes sure that
> > whattever the default toolchain are, u-boot is not impacted
> >
> > Fixes errors e.g.
> >
> > | arm-yoe-linux-gnueabi-ld.bfd: u-boot-spl section `.rodata' will not fit in region `.sram'
> > | arm-yoe-linux-gnueabi-ld.bfd: region `.sram' overflowed by 5772 bytes
> >
> > Signed-off-by: Khem Raj <raj.khem at gmail.com>
> > Cc: Tom Rini <trini at konsulko.com>
> > Cc: Denys Dmytriyenko <denys at ti.com>
> > ---
> >  .../0001-Always-build-with-fno-PIE.patch      | 35 +++++++++++++++++++
> >  .../u-boot/u-boot-ti-staging_2018.01.bb       |  2 ++
> >  2 files changed, 37 insertions(+)
> >  create mode 100644 recipes-bsp/u-boot/files/0001-Always-build-with-fno-PIE.patch
> >
> > diff --git a/recipes-bsp/u-boot/files/0001-Always-build-with-fno-PIE.patch b/recipes-bsp/u-boot/files/0001-Always-build-with-fno-PIE.patch
> > new file mode 100644
> > index 00000000..dfd99c66
> > --- /dev/null
> > +++ b/recipes-bsp/u-boot/files/0001-Always-build-with-fno-PIE.patch
> > @@ -0,0 +1,35 @@
> > +From ec81bbe3ea7e077512f4b9d2b4aa87e7f1494bcc Mon Sep 17 00:00:00 2001
> > +From: Khem Raj <raj.khem at gmail.com>
> > +Date: Wed, 17 Oct 2018 22:32:57 -0700
> > +Subject: [PATCH] Always build with -fno-PIE
> > +
> > +GNU Toolchains can be configured to generate PIE by default
> > +while PIE could be ok, it is known that u-boot/SPL on ARM starts to
> > +generate bigger code and starts to overflow the SRAM size, therefore
> > +disabling PIE explcitly insulates against toolchain intrinsic defaults
> > +
> > +Fixes errors like
> > +| arm-yoe-linux-gnueabi-ld.bfd: u-boot-spl section `.rodata' will not fit in region `.sram'
> > +| arm-yoe-linux-gnueabi-ld.bfd: region `.sram' overflowed by 5772 bytes
> > +
> > +Signed-off-by: Khem Raj <raj.khem at gmail.com>
> > +Cc: Tom Rini <trini at konsulko.com>
> > +---
> > + Makefile | 1 +
> > + 1 file changed, 1 insertion(+)
> > +
> > +diff --git a/Makefile b/Makefile
> > +index 45cd751ba3..9eb8ac564e 100644
> > +--- a/Makefile
> > ++++ b/Makefile
> > +@@ -592,6 +592,7 @@ KBUILD_CFLAGS    += -O2
> > + endif
> > +
> > + KBUILD_CFLAGS += $(call cc-option,-fno-stack-protector)
> > ++KBUILD_CFLAGS += $(call cc-option,-fno-PIE)
> > + KBUILD_CFLAGS += $(call cc-option,-fno-delete-null-pointer-checks)
> > +
> > + KBUILD_CFLAGS       += -g
> > +--
> > +2.19.1
> > +
> > diff --git a/recipes-bsp/u-boot/u-boot-ti-staging_2018.01.bb b/recipes-bsp/u-boot/u-boot-ti-staging_2018.01.bb
> > index cbb8857a..c9c97111 100644
> > --- a/recipes-bsp/u-boot/u-boot-ti-staging_2018.01.bb
> > +++ b/recipes-bsp/u-boot/u-boot-ti-staging_2018.01.bb
> > @@ -5,3 +5,5 @@ PR = "r24"
> >  BRANCH = "ti-u-boot-2018.01"
> >
> >  SRCREV = "2cc52408bf1357f11b96548e78223a1df321c1ae"
> > +
> > +SRC_URI += "file://0001-Always-build-with-fno-PIE.patch"
>
> So does this end up being something that is appropriate to push
> upstream?  Thanks!
>
> --
> Tom