i686 x PPC7450 results with 0.28-rc34

[Tom Warzeka]

I recently built a few toolchains for a PPC7450-based board using an i686
host running Red Hat Linux 9.  In order to save time with repeated attempts,
I manually downloaded all of the sources first.  These were all built with
headers from kernel version 2.4.20, because that's the version for which my
board vendor has applied patches.  Here are the results with 0.28-rc34:

GCC	GLIBC	BU	Result or error, if any (phase of failure)
-----	-----	----	--------------------------------------------------
3.2.3	2.1.3	2.14	conflicting types for `__uint128_t' (glibc)
3.2.3	2.1.3	2.15	Unrecognized opcode: `mtvrsave' (gcc-core)
3.3.4	2.1.3	2.14	missing terminating " character (glibc)
3.3.4	2.1.3	2.15	Unrecognized opcode: `mtvrsave' (gcc-core)
3.4.1	2.1.3	2.14	missing terminating " character (glibc)
3.4.1	2.1.3	2.15	missing terminating " character (glibc)

3.2.3	2.2.5	2.14	SUCCESS
3.2.3	2.2.5	2.15	Unrecognized opcode: `mtvrsave' (gcc-core)
3.3.4	2.2.5	2.14	SUCCESS
3.3.4	2.2.5	2.15	Unrecognized opcode: `mtvrsave' (gcc-core)
3.4.1	2.2.5	2.14	SUCCESS
3.4.1	2.2.5	2.15	Unrecognized opcode: `stvx' (glibc)

3.2.3	2.3.2	2.14	SUCCESS
3.2.3	2.3.2	2.15	Unrecognized opcode: `mtvrsave' (gcc-core)
3.3.4	2.3.2	2.14	SUCCESS
3.3.4	2.3.2	2.15	Unrecognized opcode: `mtvrsave' (gcc-core)
3.4.1	2.3.2	2.14	SUCCESS
3.4.1	2.3.2	2.15	SUCCESS

By applying a small patch similar to what's done in
"patches/gcc-3.4.0/gcc-3.3.3h-ppc-asm-spec.patch" to the corresponding
files in GCC 3.2.3 and 3.3.4, I was able to eliminate the "Unrecognized
opcode" errors during the build of "gcc-core".  This allowed the remaining
GLIBC 2.3.2-based toolchains to build.  Then, adding a separate patch to
GLIBC 2.2.5 similar to "patches/glibc-2.3.2/glibc-2.3.2-powerpc-as.patch"
coupled with the previous GCC patches enabled all of the GLIBC 2.2.5-based
toolchains to build.  I then used GCC-3.2.3 to successfully compile kernel
version 2.4.20.  As for verification, my application software and the
statically linked "hello" programs run fine.  I cannot try the dynamically
linked ones because my vendor provided GLIBC 2.1.3 in the RAM disk image.
Also, I have not yet tried to run the built-in tests. 

-- Tom

------
Want more information?  See the CrossGCC FAQ, http://www.objsw.com/CrossGCC/
Want to unsubscribe? Send a note to crossgcc-unsubscribe@sources.redhat.com