Aleksandar Markovic <aleksandar.m.m...@gmail.com> writes:
> On Dec 10, 2018 4:29 PM, "Alex Bennée" <alex.ben...@linaro.org> wrote: >> >> Convert the existing tests to use our common cross build >> infrastructure. >> >> [WIP: mips32r2 disabled to avoid name clash] >> [WIP: mipsr5900 disabled due to clashing build flags] >> >> Signed-off-by: Alex Bennée <alex.ben...@linaro.org> >> --- >> tests/tcg/mips/Makefile.target | 15 ++- >> tests/tcg/mips/mips32-dsp/Makefile | 166 +++++---------------------- >> tests/tcg/mips/mips32-dspr2/Makefile | 83 +++----------- >> tests/tcg/mips/mipsr5900/Makefile | 40 +++---- >> 4 files changed, 75 insertions(+), 229 deletions(-) >> > > Thank you Alex for this initiative. > > For DSP, following MIPS CPUs in QEMU support DSP: > > DSP: 24KEc (MIPS32R2), 34Kf (MIPS32R2), 74Kf (MIPS32R2), mips64dspr2 > (MIPS64R2), > > DSP R2: mips64dspr2 (MIPS64R2), That's why I'm confused by a mips32-dspr2 directory. Does that mean both of those directories run on the 74Kf processor model (MIPS32R2)? > (For simplicity, I didn't mention new nanoMIPS architecture, since your gcc > for sure don't support it yet) > > In brackets, those are architectures that are usually required to be passed > to gcc as its option, and that explains the switch involving MIPS64R2 for > DSP R2. > > Whereas while starting QEMU, CPU name should be specified in -cpu switch. > > mips64dspr2 is a virtual cpu (doesn't model any real cpu) made specifically > for DSP R2 testings, hence such odd name. > > Hopefully this clarifies some of your dilemmas. > > What are name clashes that you mention? Are they just related to > filenames? I think it is because there are a bunch of names in both mips32-dsp and mips32-dspr2 that are the same. I thought that should be sorted by the fact I only set the VPATH to tests/tcg/mips and added the tests including their dir path but for some reason it wasn't working. -- Alex Bennée
