On Fri, Nov 27, 2020 at 3:35 PM Ruifeng Wang <ruifeng.w...@arm.com> wrote: > > > -----Original Message----- > > From: David Marchand <david.march...@redhat.com> > > Sent: Friday, November 27, 2020 5:45 PM > > To: Jerin Jacob <jerinjac...@gmail.com> > > Cc: Ruifeng Wang <ruifeng.w...@arm.com>; Andrew Rybchenko > > <andrew.rybche...@oktetlabs.ru>; jer...@marvell.com; Honnappa > > Nagarahalli <honnappa.nagaraha...@arm.com>; dev <dev@dpdk.org>; nd > > <n...@arm.com>; Feifei Wang <feifei.wa...@arm.com> > > Subject: Re: [dpdk-dev] [PATCH 1/2] eal/arm: fix gcc build for optimization > > level 0 > > > > On Fri, Nov 27, 2020 at 10:34 AM Jerin Jacob <jerinjac...@gmail.com> wrote: > > > > > > On Fri, Nov 27, 2020 at 2:56 PM David Marchand > > > <david.march...@redhat.com> wrote: > > > > > > > > On Fri, Nov 27, 2020 at 9:51 AM Jerin Jacob <jerinjac...@gmail.com> > > wrote: > > > > > > > > > > On Fri, Nov 27, 2020 at 2:09 PM David Marchand > > > > > <david.march...@redhat.com> wrote: > > > > > > > > > > > > On Fri, Nov 27, 2020 at 9:10 AM Ruifeng Wang > > <ruifeng.w...@arm.com> wrote: > > > > > > > > > > > > > > Gcc build with '-O0' on platforms with > > RTE_ARM_FEATURE_ATOMICS > > > > > > > set failed for: > > > > > > > ../lib/librte_efd/rte_efd.c > > > > > > > Assembler messages: > > > > > > > 3866: Error: selected processor does not support `crc32cb > > > > > > > w0,w0,w1' > > > > > > > 3890: Error: selected processor does not support `crc32ch > > > > > > > w0,w0,w1' > > > > > > > 3914: Error: selected processor does not support `crc32cw > > w0,w0,w1' > > > > > > > 3938: Error: selected processor does not support `crc32cx > > > > > > > w0,w0,x1' > > > > > > > > > > > > > > This was caused by an architecture specifier added for Clang. > > > > > > > Unlike Clang, Gcc considers each inline assembly block to be > > > > > > > dependent and therefor, the architecture specifier impacts > > > > > > > assemble of some blocks require certain extension support. > > > > > > > > > > > > I reproduced this error with -O0 for > > config/arm/arm64_octeontx2_linux_gcc. > > > > > > The patch looks simple enough and it only affects ARM builds. > > > > > > > > > > > > Jerin, > > > > > > Are you okay with this change for 20.11? > > > > > > Or should we wait 21.02? > > > > > > > > > > I could reproduce the issue with meson build -Dbuildtype=debug > > > > > --cross-file config/arm/arm64_octeontx2_linux_gcc > > > > > > > > > > This fix is OK. Request to merge for 20.11 > > > > > > > > > > Acked-by: Jerin Jacob <jer...@marvell.com> > > > > > > > > Ruifeng, Andrew, > > > > > > > > I get a compilation error with this patch on sfc using gcc and > > > > config/arm/arm64_octeontx2_linux_gcc config + --buildtype=debug > > > > and/or > > > > -Dc_args=-O0 . > > > > > > > > aarch64-linux-gnu-gcc (GNU Toolchain for the A-profile Architecture > > > > 8.2-2019.01 (arm-rel-8.28)) 8.2.1 20180802 > > > > > > > > FAILED: > > > > drivers/common/sfc_efx/base/932d09a@@sfc_base@sta/efx_crc32.c.o > > > > aarch64-linux-gnu-gcc > > > > -Idrivers/common/sfc_efx/base/932d09a@@sfc_base@sta > > > > -Idrivers/common/sfc_efx/base > > > > -I../../dpdk/drivers/common/sfc_efx/base > > > > -Idrivers/common/sfc_efx -I../../dpdk/drivers/common/sfc_efx -I. > > > > -I../../dpdk/ -Iconfig -I../../dpdk/config -Ilib/librte_eal/include > > > > -I../../dpdk/lib/librte_eal/include -Ilib/librte_eal/linux/include > > > > -I../../dpdk/lib/librte_eal/linux/include > > > > -Ilib/librte_eal/arm/include -I../../dpdk/lib/librte_eal/arm/include > > > > -Ilib/librte_eal/common -I../../dpdk/lib/librte_eal/common > > > > -Ilib/librte_eal -I../../dpdk/lib/librte_eal -Ilib/librte_kvargs > > > > -I../../dpdk/lib/librte_kvargs > > > > -Ilib/librte_telemetry/../librte_metrics > > > > -I../../dpdk/lib/librte_telemetry/../librte_metrics > > > > -Ilib/librte_telemetry -I../../dpdk/lib/librte_telemetry > > > > -fdiagnostics-color=always -pipe -D_FILE_OFFSET_BITS=64 -Wall > > > > -Winvalid-pch -Werror -g -include rte_config.h -Wextra -Wcast-qual > > > > -Wdeprecated -Wformat -Wformat-nonliteral -Wformat-security > > > > -Wmissing-declarations -Wmissing-prototypes -Wnested-externs > > > > -Wold-style-definition -Wpointer-arith -Wsign-compare > > > > -Wstrict-prototypes -Wundef -Wwrite-strings -Wno-packed-not-aligned > > > > -Wno-missing-field-initializers -D_GNU_SOURCE -fPIC > > > > -march=armv8.2-a+crc+crypto+lse -DALLOW_EXPERIMENTAL_API > > > > -DALLOW_INTERNAL_API -Wno-format-truncation -Wdisabled- > > optimization > > > > -Waggregate-return -Wbad-function-cast -Wno-sign-compare > > > > -Wno-unused-parameter -Wno-unused-variable -Wno-empty-body > > > > -Wno-unused-but-set-variable -MD -MQ > > > > 'drivers/common/sfc_efx/base/932d09a@@sfc_base@sta/efx_crc32.c.o' > > > > -MF > > > > > > 'drivers/common/sfc_efx/base/932d09a@@sfc_base@sta/efx_crc32.c.o.d' > > > > -o > > 'drivers/common/sfc_efx/base/932d09a@@sfc_base@sta/efx_crc32.c.o' > > > > -c ../../dpdk/drivers/common/sfc_efx/base/efx_crc32.c > > > > In file included from > > > > ../../dpdk/lib/librte_eal/arm/include/rte_atomic.h:9, > > > > from > > > > ../../dpdk/lib/librte_eal/include/generic/rte_pause.h:19, > > > > from > > > > ../../dpdk/lib/librte_eal/arm/include/rte_pause_64.h:19, > > > > from > > > > ../../dpdk/lib/librte_eal/arm/include/rte_pause.h:13, > > > > from > > > > ../../dpdk/lib/librte_eal/include/generic/rte_spinlock.h:25, > > > > from > > > > ../../dpdk/lib/librte_eal/arm/include/rte_spinlock.h:17, > > > > from ../../dpdk/drivers/common/sfc_efx/efsys.h:15, > > > > from ../../dpdk/drivers/common/sfc_efx/base/efx.h:11, > > > > from > > > > ../../dpdk/drivers/common/sfc_efx/base/efx_crc32.c:7: > > > > ../../dpdk/lib/librte_eal/arm/include/rte_atomic_64.h: In function > > > > ‘__cas_128_relaxed’: > > > > ../../dpdk/lib/librte_eal/arm/include/rte_atomic_64.h:81:20: error: > > > > function returns an aggregate [-Werror=aggregate-return] > > > > __ATOMIC128_CAS_OP(__cas_128_relaxed, "casp") > > > > ^~~~~~~~~~~~~~~~~ > > > > ../../dpdk/lib/librte_eal/arm/include/rte_atomic_64.h:57:1: note: in > > > > definition of macro ‘__ATOMIC128_CAS_OP’ > > > > cas_op_name(rte_int128_t *dst, rte_int128_t old, rte_int128_t updated) > > \ > > > > ^~~~~~~~~~~ > > > > ../../dpdk/lib/librte_eal/arm/include/rte_atomic_64.h: In function > > > > ‘__cas_128_acquire’: > > > > ../../dpdk/lib/librte_eal/arm/include/rte_atomic_64.h:82:20: error: > > > > function returns an aggregate [-Werror=aggregate-return] > > > > __ATOMIC128_CAS_OP(__cas_128_acquire, "caspa") > > > > ^~~~~~~~~~~~~~~~~ > > > > ../../dpdk/lib/librte_eal/arm/include/rte_atomic_64.h:57:1: note: in > > > > definition of macro ‘__ATOMIC128_CAS_OP’ > > > > cas_op_name(rte_int128_t *dst, rte_int128_t old, rte_int128_t updated) > > \ > > > > ^~~~~~~~~~~ > > > > ../../dpdk/lib/librte_eal/arm/include/rte_atomic_64.h: In function > > > > ‘__cas_128_release’: > > > > ../../dpdk/lib/librte_eal/arm/include/rte_atomic_64.h:83:20: error: > > > > function returns an aggregate [-Werror=aggregate-return] > > > > __ATOMIC128_CAS_OP(__cas_128_release, "caspl") > > > > ^~~~~~~~~~~~~~~~~ > > > > ../../dpdk/lib/librte_eal/arm/include/rte_atomic_64.h:57:1: note: in > > > > definition of macro ‘__ATOMIC128_CAS_OP’ > > > > cas_op_name(rte_int128_t *dst, rte_int128_t old, rte_int128_t updated) > > \ > > > > ^~~~~~~~~~~ > > > > ../../dpdk/lib/librte_eal/arm/include/rte_atomic_64.h: In function > > > > ‘__cas_128_acq_rel’: > > > > ../../dpdk/lib/librte_eal/arm/include/rte_atomic_64.h:84:20: error: > > > > function returns an aggregate [-Werror=aggregate-return] > > > > __ATOMIC128_CAS_OP(__cas_128_acq_rel, "caspal") > > > > ^~~~~~~~~~~~~~~~~ > > > > ../../dpdk/lib/librte_eal/arm/include/rte_atomic_64.h:57:1: note: in > > > > definition of macro ‘__ATOMIC128_CAS_OP’ > > > > cas_op_name(rte_int128_t *dst, rte_int128_t old, rte_int128_t updated) > > \ > > > > ^~~~~~~~~~~ > > > > cc1: all warnings being treated as errors > > > > > > > > > > > > If I waive those warnings: > > > > - next to fail is net/igc (fixed by next patch so ok) > > > > - and then next to fail is net/sfc. > > > > > > Regarding the net/sfc build, Could try clean build.(I got the warning > > > first, after the clean build, it is OK) > > > > That's what I first thought too, then I noticed that test-meson-builds.sh > > had > > switched back my build dir to debugoptimized. > > > > > > From scratch: > > > > The Meson build system > > Version: 0.52.0 > > Source dir: /home/dmarchan/dpdk > > Build dir: /home/dmarchan/builds/build-arm64-octeontx2 > > Build type: cross build > > Program cat found: YES (/usr/bin/cat) > > Project name: DPDK > > Project version: 20.11.0-rc5 > > Appending CFLAGS from environment: '-Werror' > > C compiler for the build machine: ccache gcc (gcc 9.3.1 "gcc (GCC) > > 9.3.1 20200408 (Red Hat 9.3.1-2)") > > C linker for the build machine: GNU ld.bfd 2.32-33 C compiler for the host > > machine: aarch64-linux-gnu-gcc (gcc 8.2.1 "aarch64-linux-gnu-gcc (GNU > > Toolchain for the A-profile Architecture > > 8.2-2019.01 (arm-rel-8.28)) 8.2.1 20180802") C linker for the host machine: > > GNU ld.bfd 8.2-2019 Build machine cpu family: x86_64 Build machine cpu: > > x86_64 Host machine cpu family: aarch64 Host machine cpu: armv8-a Target > > machine cpu family: aarch64 Target machine cpu: armv8-a ... > > Option buildtype is: debug [default: release] Option default_library is: > > shared > > [default: static] Found ninja-1.9.0 at /usr/bin/ninja > > ninja: Entering directory `/home/dmarchan/builds/build-arm64-octeontx2' > > [479/2558] Compiling C object > > 'drivers/common/sfc_efx/base/932d09a@@sfc_base@sta/efx_bootcfg.c.o'. > > FAILED: > > drivers/common/sfc_efx/base/932d09a@@sfc_base@sta/efx_bootcfg.c.o > > aarch64-linux-gnu-gcc > > -Idrivers/common/sfc_efx/base/932d09a@@sfc_base@sta > > -Idrivers/common/sfc_efx/base -I../../dpdk/drivers/common/sfc_efx/base > > -Idrivers/common/sfc_efx -I../../dpdk/drivers/common/sfc_efx -I. > > -I../../dpdk/ -Iconfig -I../../dpdk/config -Ilib/librte_eal/include - > > I../../dpdk/lib/librte_eal/include -Ilib/librte_eal/linux/include - > > I../../dpdk/lib/librte_eal/linux/include -Ilib/librte_eal/arm/include - > > I../../dpdk/lib/librte_eal/arm/include -Ilib/librte_eal/common - > > I../../dpdk/lib/librte_eal/common -Ilib/librte_eal > > -I../../dpdk/lib/librte_eal - > > Ilib/librte_kvargs -I../../dpdk/lib/librte_kvargs - > > Ilib/librte_telemetry/../librte_metrics > > -I../../dpdk/lib/librte_telemetry/../librte_metrics > > -Ilib/librte_telemetry -I../../dpdk/lib/librte_telemetry -fdiagnostics- > > color=always -pipe -D_FILE_OFFSET_BITS=64 -Wall -Winvalid-pch -Werror -g - > > include rte_config.h -Wextra -Wcast-qual -Wdeprecated -Wformat - > > Wformat-nonliteral -Wformat-security -Wmissing-declarations -Wmissing- > > prototypes -Wnested-externs -Wold-style-definition -Wpointer-arith - > > Wsign-compare -Wstrict-prototypes -Wundef -Wwrite-strings -Wno-packed- > > not-aligned -Wno-missing-field-initializers -D_GNU_SOURCE -fPIC - > > march=armv8.2-a+crc+crypto+lse -DALLOW_EXPERIMENTAL_API - > > DALLOW_INTERNAL_API -Wno-format-truncation -Wdisabled-optimization - > > Waggregate-return -Wbad-function-cast -Wno-sign-compare -Wno-unused- > > parameter -Wno-unused-variable -Wno-empty-body -Wno-unused-but-set- > > variable -MD -MQ > > 'drivers/common/sfc_efx/base/932d09a@@sfc_base@sta/efx_bootcfg.c.o' > > -MF > > 'drivers/common/sfc_efx/base/932d09a@@sfc_base@sta/efx_bootcfg.c.o. > > d' > > -o > > 'drivers/common/sfc_efx/base/932d09a@@sfc_base@sta/efx_bootcfg.c.o' > > -c ../../dpdk/drivers/common/sfc_efx/base/efx_bootcfg.c > > In file included from ../../dpdk/lib/librte_eal/arm/include/rte_atomic.h:9, > > from > > ../../dpdk/lib/librte_eal/include/generic/rte_pause.h:19, > > from > > ../../dpdk/lib/librte_eal/arm/include/rte_pause_64.h:19, > > from ../../dpdk/lib/librte_eal/arm/include/rte_pause.h:13, > > from > > ../../dpdk/lib/librte_eal/include/generic/rte_spinlock.h:25, > > from > > ../../dpdk/lib/librte_eal/arm/include/rte_spinlock.h:17, > > from ../../dpdk/drivers/common/sfc_efx/efsys.h:15, > > from ../../dpdk/drivers/common/sfc_efx/base/efx.h:11, > > from > > ../../dpdk/drivers/common/sfc_efx/base/efx_bootcfg.c:7: > > ../../dpdk/lib/librte_eal/arm/include/rte_atomic_64.h: In function > > ‘__cas_128_relaxed’: > > ../../dpdk/lib/librte_eal/arm/include/rte_atomic_64.h:81:20: error: > > function returns an aggregate [-Werror=aggregate-return] > > __ATOMIC128_CAS_OP(__cas_128_relaxed, "casp") ... > > I can see similar warning message (function returns an aggregate) when doing > "-O0" build. > Since "-Werror=aggregate-return" is used, error will be reported. > This is because '__cas_128_xxx' return aggregated data type I think. > Inline the functions should resolve the issue. But it could cause register > corruption as discussed > when the code was added: > http://patches.dpdk.org/patch/55580/ > > No quick solution I can think of.
Yes. We may need to return two separate uint64_t by taking as the pointer. It will be NOT a candidate for 20.11. In my setup: meson --buildtype=debug build --cross-file config/arm/arm64_octeontx2_linux_gcc ninja -C build Shows the warnings for "warning: function returns an aggregate [-Waggregate-return]", Since warning not treated as the error it links. I think, for this release, The current[1] minor changes are OK as it improves the situation. http://patches.dpdk.org/project/dpdk/list/?submitter=Ruifeng > > > > > > -- > > David Marchand >
