Message ID | 20220331145813.21719-1-leandro.lupori@eldorado.org.br (mailing list archive) |
---|---|
Headers | show |
Series | Port PPC64/PowerNV MMU tests to QEMU | expand |
Hello Leandro, I think we should start looking at semihosting like Richard proposed and for it, rework the QEMU attn implementation or may be use a special form of scv. It should not change too much the tests, only exit and output which would remove the dependency on the platform console. In the long run, it looks like a good investment for more low level tests. Thanks, C. On 3/31/22 16:58, Leandro Lupori wrote: > Changes from v1: > - added new files to MAINTAINERS, under PowerPC TCG CPUs, and added > myself as a reviewer (not sure if this is the right procedure) > - removed the "virtual" ppc64le-softmmu target. > ppc64/Makefile.softmmu-target now builds all tests in both BE and LE > > Cédric Le Goater (2): > target/ppc: Add support for the Processor Attention instruction > ppc/pnv: Activate support for the Processor Attention instruction > > Leandro Lupori (3): > tests/tcg/ppc64: add basic softmmu test support > tests/tcg/ppc64: add MMU test sources > tests/tcg/ppc64: build PowerNV and LE tests > > MAINTAINERS | 2 + > hw/ppc/pnv_core.c | 6 + > include/hw/ppc/pnv_core.h | 1 + > target/ppc/cpu.h | 8 + > target/ppc/excp_helper.c | 27 + > target/ppc/helper.h | 1 + > target/ppc/translate.c | 14 + > tests/tcg/ppc64/Makefile.softmmu-rules | 34 + > tests/tcg/ppc64/Makefile.softmmu-target | 124 ++++ > tests/tcg/ppc64/system/include/asm.h | 62 ++ > tests/tcg/ppc64/system/include/console.h | 15 + > tests/tcg/ppc64/system/include/io.h | 61 ++ > tests/tcg/ppc64/system/include/pnv.h | 21 + > tests/tcg/ppc64/system/include/uart.h | 54 ++ > tests/tcg/ppc64/system/lib/boot.S | 68 ++ > tests/tcg/ppc64/system/lib/console.c | 173 +++++ > tests/tcg/ppc64/system/lib/powerpc.lds | 27 + > tests/tcg/ppc64/system/mmu-head.S | 142 +++++ > tests/tcg/ppc64/system/mmu.c | 764 +++++++++++++++++++++++ > tests/tcg/ppc64/system/mmu.h | 9 + > 20 files changed, 1613 insertions(+) > create mode 100644 tests/tcg/ppc64/Makefile.softmmu-rules > create mode 100644 tests/tcg/ppc64/Makefile.softmmu-target > create mode 100644 tests/tcg/ppc64/system/include/asm.h > create mode 100644 tests/tcg/ppc64/system/include/console.h > create mode 100644 tests/tcg/ppc64/system/include/io.h > create mode 100644 tests/tcg/ppc64/system/include/pnv.h > create mode 100644 tests/tcg/ppc64/system/include/uart.h > create mode 100644 tests/tcg/ppc64/system/lib/boot.S > create mode 100644 tests/tcg/ppc64/system/lib/console.c > create mode 100644 tests/tcg/ppc64/system/lib/powerpc.lds > create mode 100644 tests/tcg/ppc64/system/mmu-head.S > create mode 100644 tests/tcg/ppc64/system/mmu.c > create mode 100644 tests/tcg/ppc64/system/mmu.h >
On 3/31/22 12:47, Cédric Le Goater wrote: > > Hello Leandro, > > I think we should start looking at semihosting like Richard proposed > and for it, rework the QEMU attn implementation or may be use a special > form of scv. > > It should not change too much the tests, only exit and output which > would remove the dependency on the platform console. In the long run, > it looks like a good investment for more low level tests. > > Thanks, > > C. > Hi Cédric, I've assumed you would handle the comments related to the attn instruction implementation, so that's why I've sent a v2 with only the other changes. But no problem, we can coordinate how we're going to add semihosting support to PPC. Thanks, Leandro