| Message ID | 20190430165234.32272-1-alex.bennee@linaro.org |
|---|---|
| Headers | show |
| Series | demacro softmmu (plus tests/coverage) | expand |
On 4/30/19 9:52 AM, Alex Bennée wrote: > I've also moved the existing system memory test and made it multiarch > and added the bootstrapping for aarch64 system tests. I would like to > add support for Big Endian as well but I didn't want to delay the > posting of the series. It would also be nice to exercise the > ioread/write paths and other handling but I leave this as an exercise > for later. Somewhere in there you're adding -chardev file,path=hello.out,id=output but there's no corresponding use of the chardev. Which, somehow doesn't seem to matter to your aarch64 testcase, but when I try this for alpha I truly get no output at all. I needed -serial chardev:output to populate the file. r~
Richard Henderson <richard.henderson@linaro.org> writes: > On 4/30/19 9:52 AM, Alex Bennée wrote: >> I've also moved the existing system memory test and made it multiarch >> and added the bootstrapping for aarch64 system tests. I would like to >> add support for Big Endian as well but I didn't want to delay the >> posting of the series. It would also be nice to exercise the >> ioread/write paths and other handling but I leave this as an exercise >> for later. > > Somewhere in there you're adding > > -chardev file,path=hello.out,id=output It's in the default runner config in tests/tcg/Makefile > but there's no corresponding use of the chardev. > Which, somehow doesn't seem to matter to your aarch64 > testcase, Argh.. it's because -semihosting is different from -serial and doesn't allow the usual redirection rules you get with a chardev.. > but when I try this for alpha I truly get > no output at all. I needed > > -serial chardev:output or -serial chadev,id=output? > > to populate the file. > > > r~ -- Alex Bennée
Alex Bennée <alex.bennee@linaro.org> writes: > Hi, > > This is the latest iteration of the softmmu demacro series. The main > changes from the last submission are some updates from Richard. Ping Emilio/Mark Would you be able to re-run your tests to check there are no other regressions? I can then get the PR prepared for merging ;-) > Some > are merged into the main demacro patch (mostly casting cleanups) and > then a series of 3 patches to out of line some of the less hot code to > keep the main access functions a reasonable size. For example: > > readelf -s aarch64-softmmu/accel/tcg/cputlb.o | ag helper |\ > ag -v atomic | ag mmu | tr -s ' ' | cut -d ' ' -f 4,9 | sort -n > > Before: > > 16 helper_be_ldsl_mmu > 16 helper_le_ldsl_mmu > 18 helper_be_ldsw_mmu > 18 helper_le_ldsw_mmu > 18 helper_ret_ldsb_mmu > 535 helper_ret_ldub_mmu > 556 helper_ret_ldb_cmmu > 586 helper_ret_stb_mmu > 679 helper_le_ldul_mmu > 681 helper_be_ldul_mmu > 685 helper_le_ldq_mmu > 688 helper_be_ldq_mmu > 688 helper_le_lduw_mmu > 693 helper_le_ldl_cmmu > 701 helper_le_ldq_cmmu > 701 helper_le_ldw_cmmu > 703 helper_be_ldl_cmmu > 704 helper_be_ldq_cmmu > 708 helper_be_lduw_mmu > 713 helper_be_ldw_cmmu > 943 helper_le_stw_mmu > 944 helper_le_stl_mmu > 952 helper_be_stl_mmu > 952 helper_le_stq_mmu > 959 helper_be_stw_mmu > 960 helper_be_stq_mmu > > After: > > 5 helper_be_ldul_mmu > 5 helper_be_lduw_mmu > 5 helper_le_ldul_mmu > 5 helper_le_lduw_mmu > 5 helper_ret_ldub_mmu > 14 helper_be_ldl_cmmu > 14 helper_be_ldw_cmmu > 14 helper_le_ldl_cmmu > 14 helper_le_ldw_cmmu > 16 helper_be_ldsl_mmu > 16 helper_le_ldsl_mmu > 18 helper_be_ldsw_mmu > 18 helper_le_ldsw_mmu > 18 helper_ret_ldsb_mmu > 783 helper_ret_stb_mmu > 785 helper_ret_ldb_cmmu > 881 helper_be_ldq_mmu > 881 helper_le_ldq_mmu > 889 helper_le_ldq_cmmu > 897 helper_be_ldq_cmmu > 1150 helper_be_stw_mmu > 1150 helper_le_stw_mmu > 1151 helper_be_stq_mmu > 1151 helper_le_stl_mmu > 1151 helper_le_stq_mmu > 1159 helper_be_stl_mmu > > I've also moved the existing system memory test and made it multiarch > and added the bootstrapping for aarch64 system tests. I would like to > add support for Big Endian as well but I didn't want to delay the > posting of the series. It would also be nice to exercise the > ioread/write paths and other handling but I leave this as an exercise > for later. > > There are also some minor tweaks for the code coverage reports now I'm > running with out-of-tree builds. > > Alex Bennée (11): > tests/tcg/multiarch: add support for multiarch system tests > tests/tcg/multiarch: add hello world system test > tests/tcg/aarch64: add system boot.S > tests/tcg/multiarch: move the system memory test > tests/tcg/minilib: support %c format char > tests/tcg/multiarch: expand system memory test to cover more > accel/tcg: demacro cputlb > accel/tcg: remove softmmu_template.h > Makefile: fix coverage-report reference to BUILD_DIR > Makefile: include per-target build directories in coverage report > Makefile.target: support per-target coverage reports > > Richard Henderson (4): > cputlb: Move TLB_RECHECK handling into load/store_helper > cputlb: Drop attribute flatten > cputlb: Do unaligned load recursion to outermost function > cputlb: Do unaligned store recursion to outermost function > > Makefile | 4 +- > Makefile.target | 16 + > accel/tcg/cputlb.c | 626 +++++++++++++++--- > accel/tcg/softmmu_template.h | 454 ------------- > tests/tcg/Makefile | 1 + > tests/tcg/aarch64/Makefile.softmmu-target | 32 + > tests/tcg/aarch64/system/boot.S | 200 ++++++ > tests/tcg/aarch64/system/kernel.ld | 22 + > tests/tcg/i386/Makefile.softmmu-target | 2 +- > tests/tcg/i386/system/memory.c | 243 ------- > tests/tcg/minilib/printf.c | 6 +- > .../multiarch/system/Makefile.softmmu-target | 14 + > tests/tcg/{i386 => multiarch}/system/hello.c | 0 > tests/tcg/multiarch/system/memory.c | 427 ++++++++++++ > 14 files changed, 1267 insertions(+), 780 deletions(-) > delete mode 100644 accel/tcg/softmmu_template.h > create mode 100644 tests/tcg/aarch64/Makefile.softmmu-target > create mode 100644 tests/tcg/aarch64/system/boot.S > create mode 100644 tests/tcg/aarch64/system/kernel.ld > delete mode 100644 tests/tcg/i386/system/memory.c > create mode 100644 tests/tcg/multiarch/system/Makefile.softmmu-target > rename tests/tcg/{i386 => multiarch}/system/hello.c (100%) > create mode 100644 tests/tcg/multiarch/system/memory.c -- Alex Bennée
On 10/05/2019 11:36, Alex Bennée wrote: > Alex Bennée <alex.bennee@linaro.org> writes: > >> Hi, >> >> This is the latest iteration of the softmmu demacro series. The main >> changes from the last submission are some updates from Richard. > > Ping Emilio/Mark > > Would you be able to re-run your tests to check there are no other > regressions? I can then get the PR prepared for merging ;-) Hi Alex, Sorry for the delay - I've run this through my complete set of PPC/SPARC32/SPARC64 test images and no regressions as far as I can see with this latest version. Tested-by: Mark Cave-Ayland <mark.cave-ayland@ilande.co.uk> ATB, Mark.
On Fri, May 10, 2019 at 11:36:33 +0100, Alex Bennée wrote: > Ping Emilio/Mark > > Would you be able to re-run your tests to check there are no other > regressions? I can then get the PR prepared for merging ;-) I'll try to run some tests next week, but I am not sure I'll have time to do so. If I were you I'd go ahead with the PR -- it's best to have these type of changes merged early in the development cycle. Thanks, Emilio
Hi, This is the latest iteration of the softmmu demacro series. The main changes from the last submission are some updates from Richard. Some are merged into the main demacro patch (mostly casting cleanups) and then a series of 3 patches to out of line some of the less hot code to keep the main access functions a reasonable size. For example: readelf -s aarch64-softmmu/accel/tcg/cputlb.o | ag helper |\ ag -v atomic | ag mmu | tr -s ' ' | cut -d ' ' -f 4,9 | sort -n Before: 16 helper_be_ldsl_mmu 16 helper_le_ldsl_mmu 18 helper_be_ldsw_mmu 18 helper_le_ldsw_mmu 18 helper_ret_ldsb_mmu 535 helper_ret_ldub_mmu 556 helper_ret_ldb_cmmu 586 helper_ret_stb_mmu 679 helper_le_ldul_mmu 681 helper_be_ldul_mmu 685 helper_le_ldq_mmu 688 helper_be_ldq_mmu 688 helper_le_lduw_mmu 693 helper_le_ldl_cmmu 701 helper_le_ldq_cmmu 701 helper_le_ldw_cmmu 703 helper_be_ldl_cmmu 704 helper_be_ldq_cmmu 708 helper_be_lduw_mmu 713 helper_be_ldw_cmmu 943 helper_le_stw_mmu 944 helper_le_stl_mmu 952 helper_be_stl_mmu 952 helper_le_stq_mmu 959 helper_be_stw_mmu 960 helper_be_stq_mmu After: 5 helper_be_ldul_mmu 5 helper_be_lduw_mmu 5 helper_le_ldul_mmu 5 helper_le_lduw_mmu 5 helper_ret_ldub_mmu 14 helper_be_ldl_cmmu 14 helper_be_ldw_cmmu 14 helper_le_ldl_cmmu 14 helper_le_ldw_cmmu 16 helper_be_ldsl_mmu 16 helper_le_ldsl_mmu 18 helper_be_ldsw_mmu 18 helper_le_ldsw_mmu 18 helper_ret_ldsb_mmu 783 helper_ret_stb_mmu 785 helper_ret_ldb_cmmu 881 helper_be_ldq_mmu 881 helper_le_ldq_mmu 889 helper_le_ldq_cmmu 897 helper_be_ldq_cmmu 1150 helper_be_stw_mmu 1150 helper_le_stw_mmu 1151 helper_be_stq_mmu 1151 helper_le_stl_mmu 1151 helper_le_stq_mmu 1159 helper_be_stl_mmu I've also moved the existing system memory test and made it multiarch and added the bootstrapping for aarch64 system tests. I would like to add support for Big Endian as well but I didn't want to delay the posting of the series. It would also be nice to exercise the ioread/write paths and other handling but I leave this as an exercise for later. There are also some minor tweaks for the code coverage reports now I'm running with out-of-tree builds. Alex Bennée (11): tests/tcg/multiarch: add support for multiarch system tests tests/tcg/multiarch: add hello world system test tests/tcg/aarch64: add system boot.S tests/tcg/multiarch: move the system memory test tests/tcg/minilib: support %c format char tests/tcg/multiarch: expand system memory test to cover more accel/tcg: demacro cputlb accel/tcg: remove softmmu_template.h Makefile: fix coverage-report reference to BUILD_DIR Makefile: include per-target build directories in coverage report Makefile.target: support per-target coverage reports Richard Henderson (4): cputlb: Move TLB_RECHECK handling into load/store_helper cputlb: Drop attribute flatten cputlb: Do unaligned load recursion to outermost function cputlb: Do unaligned store recursion to outermost function Makefile | 4 +- Makefile.target | 16 + accel/tcg/cputlb.c | 626 +++++++++++++++--- accel/tcg/softmmu_template.h | 454 ------------- tests/tcg/Makefile | 1 + tests/tcg/aarch64/Makefile.softmmu-target | 32 + tests/tcg/aarch64/system/boot.S | 200 ++++++ tests/tcg/aarch64/system/kernel.ld | 22 + tests/tcg/i386/Makefile.softmmu-target | 2 +- tests/tcg/i386/system/memory.c | 243 ------- tests/tcg/minilib/printf.c | 6 +- .../multiarch/system/Makefile.softmmu-target | 14 + tests/tcg/{i386 => multiarch}/system/hello.c | 0 tests/tcg/multiarch/system/memory.c | 427 ++++++++++++ 14 files changed, 1267 insertions(+), 780 deletions(-) delete mode 100644 accel/tcg/softmmu_template.h create mode 100644 tests/tcg/aarch64/Makefile.softmmu-target create mode 100644 tests/tcg/aarch64/system/boot.S create mode 100644 tests/tcg/aarch64/system/kernel.ld delete mode 100644 tests/tcg/i386/system/memory.c create mode 100644 tests/tcg/multiarch/system/Makefile.softmmu-target rename tests/tcg/{i386 => multiarch}/system/hello.c (100%) create mode 100644 tests/tcg/multiarch/system/memory.c -- 2.20.1