| Message ID | 20200610203942.887374-6-richard.henderson@linaro.org |
|---|---|
| State | New |
| Headers | show |
| Series | Vs clang-10 and gcc-9 warnings | expand |
Richard Henderson <richard.henderson@linaro.org> writes: > On aarch64, gcc 9.3 is generating > > qemu/exec.c: In function ‘address_space_translate_iommu’: > qemu/exec.c:431:28: note: parameter passing for argument of type \ > ‘MemTxAttrs’ {aka ‘struct MemTxAttrs’} changed in GCC 9.1 > > and many other reptitions. This structure, and the functions > amongst which it is passed, are not part of a QEMU public API. > Therefore we do not care how the compiler passes the argument, > so long as the compiler is self-consistent. > > Cc: Alex Bennée <alex.bennee@linaro.org> > Cc: Peter Maydell <peter.maydell@linaro.org> > Signed-off-by: Richard Henderson <richard.henderson@linaro.org> > --- > TODO: The only portion of QEMU which does have a public api, > and so must have a stable abi, is "qemu/plugin.h". We could > test this by forcing -Wpsabi or -Werror=psabi in tests/plugin. > I can't seem to make that work -- Alex? modified plugins/Makefile.objs @@ -5,6 +5,7 @@ obj-y += loader.o obj-y += core.o obj-y += api.o +api.o-cflags := -Wpsabi # Abuse -libs suffix to only link with --dynamic-list/-exported_symbols_list # when the final binary includes the plugin object. Seems to work for me. -- Alex Bennée
On 6/11/20 9:44 AM, Alex Bennée wrote: > > Richard Henderson <richard.henderson@linaro.org> writes: > >> On aarch64, gcc 9.3 is generating >> >> qemu/exec.c: In function ‘address_space_translate_iommu’: >> qemu/exec.c:431:28: note: parameter passing for argument of type \ >> ‘MemTxAttrs’ {aka ‘struct MemTxAttrs’} changed in GCC 9.1 >> >> and many other reptitions. This structure, and the functions >> amongst which it is passed, are not part of a QEMU public API. >> Therefore we do not care how the compiler passes the argument, >> so long as the compiler is self-consistent. >> >> Cc: Alex Bennée <alex.bennee@linaro.org> >> Cc: Peter Maydell <peter.maydell@linaro.org> >> Signed-off-by: Richard Henderson <richard.henderson@linaro.org> >> --- >> TODO: The only portion of QEMU which does have a public api, >> and so must have a stable abi, is "qemu/plugin.h". We could >> test this by forcing -Wpsabi or -Werror=psabi in tests/plugin. >> I can't seem to make that work -- Alex? > > modified plugins/Makefile.objs > @@ -5,6 +5,7 @@ > obj-y += loader.o > obj-y += core.o > obj-y += api.o > +api.o-cflags := -Wpsabi > > # Abuse -libs suffix to only link with --dynamic-list/-exported_symbols_list > # when the final binary includes the plugin object. > > Seems to work for me. Wrong directory -- that's the part that goes into qemu, which also uses other qemu internal headers. As opposed to the tests/, which only use the one "qemu/plugins.h" header (plus libc). r~
Richard Henderson <richard.henderson@linaro.org> writes: > On 6/11/20 9:44 AM, Alex Bennée wrote: >> >> Richard Henderson <richard.henderson@linaro.org> writes: >> >>> On aarch64, gcc 9.3 is generating >>> >>> qemu/exec.c: In function ‘address_space_translate_iommu’: >>> qemu/exec.c:431:28: note: parameter passing for argument of type \ >>> ‘MemTxAttrs’ {aka ‘struct MemTxAttrs’} changed in GCC 9.1 >>> >>> and many other reptitions. This structure, and the functions >>> amongst which it is passed, are not part of a QEMU public API. >>> Therefore we do not care how the compiler passes the argument, >>> so long as the compiler is self-consistent. >>> >>> Cc: Alex Bennée <alex.bennee@linaro.org> >>> Cc: Peter Maydell <peter.maydell@linaro.org> >>> Signed-off-by: Richard Henderson <richard.henderson@linaro.org> >>> --- >>> TODO: The only portion of QEMU which does have a public api, >>> and so must have a stable abi, is "qemu/plugin.h". We could >>> test this by forcing -Wpsabi or -Werror=psabi in tests/plugin. >>> I can't seem to make that work -- Alex? >> >> modified plugins/Makefile.objs >> @@ -5,6 +5,7 @@ >> obj-y += loader.o >> obj-y += core.o >> obj-y += api.o >> +api.o-cflags := -Wpsabi >> >> # Abuse -libs suffix to only link with --dynamic-list/-exported_symbols_list >> # when the final binary includes the plugin object. >> >> Seems to work for me. > > Wrong directory -- that's the part that goes into qemu, which also uses other > qemu internal headers. As opposed to the tests/, which only use the one > "qemu/plugins.h" header (plus libc). It's a sub-make so I just did: modified tests/plugin/Makefile @@ -18,7 +18,7 @@ NAMES += hwprofile SONAMES := $(addsuffix .so,$(addprefix lib,$(NAMES))) -QEMU_CFLAGS += -fPIC +QEMU_CFLAGS += -fPIC -Wpsabi QEMU_CFLAGS += -I$(SRC_PATH)/include/qemu -- Alex Bennée
On 6/11/20 10:17 AM, Alex Bennée wrote: > > Richard Henderson <richard.henderson@linaro.org> writes: > >> On 6/11/20 9:44 AM, Alex Bennée wrote: >>> >>> Richard Henderson <richard.henderson@linaro.org> writes: >>> >>>> On aarch64, gcc 9.3 is generating >>>> >>>> qemu/exec.c: In function ‘address_space_translate_iommu’: >>>> qemu/exec.c:431:28: note: parameter passing for argument of type \ >>>> ‘MemTxAttrs’ {aka ‘struct MemTxAttrs’} changed in GCC 9.1 >>>> >>>> and many other reptitions. This structure, and the functions >>>> amongst which it is passed, are not part of a QEMU public API. >>>> Therefore we do not care how the compiler passes the argument, >>>> so long as the compiler is self-consistent. >>>> >>>> Cc: Alex Bennée <alex.bennee@linaro.org> >>>> Cc: Peter Maydell <peter.maydell@linaro.org> >>>> Signed-off-by: Richard Henderson <richard.henderson@linaro.org> >>>> --- >>>> TODO: The only portion of QEMU which does have a public api, >>>> and so must have a stable abi, is "qemu/plugin.h". We could >>>> test this by forcing -Wpsabi or -Werror=psabi in tests/plugin. >>>> I can't seem to make that work -- Alex? >>> >>> modified plugins/Makefile.objs >>> @@ -5,6 +5,7 @@ >>> obj-y += loader.o >>> obj-y += core.o >>> obj-y += api.o >>> +api.o-cflags := -Wpsabi >>> >>> # Abuse -libs suffix to only link with --dynamic-list/-exported_symbols_list >>> # when the final binary includes the plugin object. >>> >>> Seems to work for me. >> >> Wrong directory -- that's the part that goes into qemu, which also uses other >> qemu internal headers. As opposed to the tests/, which only use the one >> "qemu/plugins.h" header (plus libc). > > It's a sub-make so I just did: > > modified tests/plugin/Makefile > @@ -18,7 +18,7 @@ NAMES += hwprofile > > SONAMES := $(addsuffix .so,$(addprefix lib,$(NAMES))) > > -QEMU_CFLAGS += -fPIC > +QEMU_CFLAGS += -fPIC -Wpsabi > QEMU_CFLAGS += -I$(SRC_PATH)/include/qemu Did you look at the actual flags passed to the actual cc via V=1? Neither of these flags is arriving. I sent you mail about this yesterday... r~
Richard Henderson <richard.henderson@linaro.org> writes: > On 6/11/20 10:17 AM, Alex Bennée wrote: >> >> Richard Henderson <richard.henderson@linaro.org> writes: >> >>> On 6/11/20 9:44 AM, Alex Bennée wrote: >>>> >>>> Richard Henderson <richard.henderson@linaro.org> writes: >>>> >>>>> On aarch64, gcc 9.3 is generating >>>>> >>>>> qemu/exec.c: In function ‘address_space_translate_iommu’: >>>>> qemu/exec.c:431:28: note: parameter passing for argument of type \ >>>>> ‘MemTxAttrs’ {aka ‘struct MemTxAttrs’} changed in GCC 9.1 >>>>> >>>>> and many other reptitions. This structure, and the functions >>>>> amongst which it is passed, are not part of a QEMU public API. >>>>> Therefore we do not care how the compiler passes the argument, >>>>> so long as the compiler is self-consistent. >>>>> >>>>> Cc: Alex Bennée <alex.bennee@linaro.org> >>>>> Cc: Peter Maydell <peter.maydell@linaro.org> >>>>> Signed-off-by: Richard Henderson <richard.henderson@linaro.org> >>>>> --- >>>>> TODO: The only portion of QEMU which does have a public api, >>>>> and so must have a stable abi, is "qemu/plugin.h". We could >>>>> test this by forcing -Wpsabi or -Werror=psabi in tests/plugin. >>>>> I can't seem to make that work -- Alex? >>>> >>>> modified plugins/Makefile.objs >>>> @@ -5,6 +5,7 @@ >>>> obj-y += loader.o >>>> obj-y += core.o >>>> obj-y += api.o >>>> +api.o-cflags := -Wpsabi >>>> >>>> # Abuse -libs suffix to only link with --dynamic-list/-exported_symbols_list >>>> # when the final binary includes the plugin object. >>>> >>>> Seems to work for me. >>> >>> Wrong directory -- that's the part that goes into qemu, which also uses other >>> qemu internal headers. As opposed to the tests/, which only use the one >>> "qemu/plugins.h" header (plus libc). >> >> It's a sub-make so I just did: >> >> modified tests/plugin/Makefile >> @@ -18,7 +18,7 @@ NAMES += hwprofile >> >> SONAMES := $(addsuffix .so,$(addprefix lib,$(NAMES))) >> >> -QEMU_CFLAGS += -fPIC >> +QEMU_CFLAGS += -fPIC -Wpsabi >> QEMU_CFLAGS += -I$(SRC_PATH)/include/qemu > > Did you look at the actual flags passed to the actual cc via V=1? > Neither of these flags is arriving. I did: cc -iquote /home/alex/lsrc/qemu.git/builds/all.plugin/. -iquote . -iquote /home/alex/lsrc/qemu.git/tcg/i386 -isystem /home/alex/lsrc/qemu.git/linux-headers -isystem /home/alex/lsrc/qemu.git/builds/all.plugin/linux-headers -iquote . -iquote /home/alex/lsrc/qemu.git -iquote /home/alex/lsrc/qemu.git/accel/tcg -iquote /home/alex/lsrc/qemu.git/include -iquote /home/alex/lsrc/qemu.git/disas/libvixl -I/usr/include/pixman-1 -I/home/alex/lsrc/qemu.git/dtc/libfdt -Werror -pthread -I/usr/include/glib-2.0 -I/usr/lib/x86_64-linux-gnu/glib-2.0/include -pthread -I/usr/include/glib-2.0 -I/usr/lib/x86_64-linux-gnu/glib-2.0/include -fPIE -DPIE -m64 -mcx16 -D_GNU_SOURCE -D_FILE_OFFSET_BITS=64 -D_LARGEFILE_SOURCE -Wstrict-prototypes -Wredundant-decls -Wall -Wundef -Wwrite-strings -Wmissing-prototypes -fno-strict-aliasing -fno-common -fwrapv -std=gnu99 -Og -ggdb3 -fvar-tracking-assignments -Wexpansion-to-defined -Wendif-labels -Wno-shift-negative-value -Wno-missing-include-dirs -Wempty-body -Wnested-externs -Wformat-security -Wformat-y2k -Winit-self -Wignored-qualifiers -Wold-style-declaration -Wold-style-definition -Wtype-limits -fstack-protector-strong -I/usr/include/p11-kit-1 -DLEGACY_RDMA_REG_MR -DSTRUCT_IOVEC_DEFINED -I/usr/include/libpng16 -I/usr/include/libdrm -I/usr/include/spice-server -I/usr/include/spice-1 -I/usr/include/capstone -fPIC -Wpsabi -I/home/alex/lsrc/qemu.git/include/qemu -MMD -MP -MT hwprofile.o -MF ./hwprofile.d -O2 -U_FORTIFY_SOURCE -D_FORTIFY_SOURCE=2 -g -c -o hwprofile.o /home/alex/lsrc/qemu.git/tests/plugin/hwprofile.c It's nested between the -I's with -fPIC. > > I sent you mail about this yesterday... > > > r~ -- Alex Bennée
diff --git a/configure b/configure index 8b33447048..76d32e0f7b 100755 --- a/configure +++ b/configure @@ -2036,6 +2036,7 @@ add_to nowarn_flags -Wno-shift-negative-value add_to nowarn_flags -Wno-string-plus-int add_to nowarn_flags -Wno-typedef-redefinition add_to nowarn_flags -Wno-tautological-type-limit-compare +add_to nowarn_flags -Wno-psabi gcc_flags="$warn_flags $nowarn_flags"
On aarch64, gcc 9.3 is generating qemu/exec.c: In function ‘address_space_translate_iommu’: qemu/exec.c:431:28: note: parameter passing for argument of type \ ‘MemTxAttrs’ {aka ‘struct MemTxAttrs’} changed in GCC 9.1 and many other reptitions. This structure, and the functions amongst which it is passed, are not part of a QEMU public API. Therefore we do not care how the compiler passes the argument, so long as the compiler is self-consistent. Cc: Alex Bennée <alex.bennee@linaro.org> Cc: Peter Maydell <peter.maydell@linaro.org> Signed-off-by: Richard Henderson <richard.henderson@linaro.org> --- TODO: The only portion of QEMU which does have a public api, and so must have a stable abi, is "qemu/plugin.h". We could test this by forcing -Wpsabi or -Werror=psabi in tests/plugin. I can't seem to make that work -- Alex? --- configure | 1 + 1 file changed, 1 insertion(+) -- 2.25.1