Re: [PATCH v4] Run tests only if the machine supports the instruction set.
> gcc/ChangeLog-archlevel > > * config/s390/s390-c.c (s390_cpu_cpp_builtins_internal): Define > __S390_ARCH_LEVEL__. > gcc/testsuite/ChangeLog-setmem > > * gcc.target/s390/md/setmem_long-1.c: Use "s390_useable_hw". > * gcc.target/s390/md/rXsbg_mode_sXl.c: Likewise. > * gcc.target/s390/md/andc-splitter-1.c: Likewise. > * gcc.target/s390/md/andc-splitter-2.c: Likewise. > * lib/gcc-dg.exp (gcc-dg-runtest): Export torture_current_flags. > * gcc.target/s390/s390.exp: Import torture_current_flags. > (check_effective_target_s390_useable_hw): New. > (check_effective_target_s390_z900_hw): New. > (check_effective_target_s390_z990_hw): New. > (check_effective_target_s390_z9_ec_hw): New. > (check_effective_target_s390_z10_hw): New. > (check_effective_target_s390_z196_hw): New. > (check_effective_target_s390_zEC12_hw): New. > (check_effective_target_s390_z13_hw): New. > (check_effective_target_z10_instructions): Removed. > (torture tests): Add optimization level without -march=. > Reorder torture tests for good cache usage. Applied. Thanks! -Andreas-
Re: [PATCH v4] Run tests only if the machine supports the instruction set.
On Tue, Dec 20, 2016 at 10:32:26PM +0100, Jakub Jelinek wrote: > On Tue, Dec 20, 2016 at 10:26:13PM +0100, Dominik Vogt wrote: > > On Tue, Dec 20, 2016 at 11:57:52AM -0800, Mike Stump wrote: > > > On Dec 20, 2016, at 6:10 AM, Dominik Vogtwrote: > > > > Right, it gets called even more often than one would think, and > > > > even with empty torture_current_options. The attached new patch > > > > (v3) removes -Ox options and superflous whitespace and caches that > > > > between calls if it's not empty. There's another, permanent cache > > > > for calls without any flags. With proper ordering of the torture > > > > options, the test program is built only a couple of times. > > > > > > Seems fine to me, but most other cases use the postfix _hw. Any > > > reason not use use _hw (and not _runable) on these? If not, > > > could you please use _hw instead. > > > > No specific reason other than lack of imagination. "s390_hw" is a > > bit too generic in my eyes -> the new names are: > > > > v4: > > > > * Renamed "s390_runnable" to "s390_useable_hw". > > * Renamed "z900_runnable" to "s390_z900_hw", > > Renamed "z10_runnable" to "s390_z10_hw", > > etc. > > Grepping for _hw in target-supports.exp reveals that usually the > effective target predicates are called _hw or _hw_available, > __hw only if it is too ambiguous (e.g. alpha_max_hw or > ppc_float128_hw_available). So I think z900_hw, z10_hw etc. is good > enough (as long as it does not clash with some other target isa name), > s390_usable_hw or s390_hw_available is fine. Okay. We usually prefix everyting with "s390_" on S/390, so I'd say we don't make an exception here - even if there are no potential naming collisions. Ciao Dominik ^_^ ^_^ -- Dominik Vogt IBM Germany
Re: [PATCH v4] Run tests only if the machine supports the instruction set.
On Tue, Dec 20, 2016 at 10:26:13PM +0100, Dominik Vogt wrote: > On Tue, Dec 20, 2016 at 11:57:52AM -0800, Mike Stump wrote: > > On Dec 20, 2016, at 6:10 AM, Dominik Vogtwrote: > > > Right, it gets called even more often than one would think, and > > > even with empty torture_current_options. The attached new patch > > > (v3) removes -Ox options and superflous whitespace and caches that > > > between calls if it's not empty. There's another, permanent cache > > > for calls without any flags. With proper ordering of the torture > > > options, the test program is built only a couple of times. > > > > Seems fine to me, but most other cases use the postfix _hw. Any > > reason not use use _hw (and not _runable) on these? If not, > > could you please use _hw instead. > > No specific reason other than lack of imagination. "s390_hw" is a > bit too generic in my eyes -> the new names are: > > v4: > > * Renamed "s390_runnable" to "s390_useable_hw". > * Renamed "z900_runnable" to "s390_z900_hw", > Renamed "z10_runnable" to "s390_z10_hw", > etc. Grepping for _hw in target-supports.exp reveals that usually the effective target predicates are called _hw or _hw_available, __hw only if it is too ambiguous (e.g. alpha_max_hw or ppc_float128_hw_available). So I think z900_hw, z10_hw etc. is good enough (as long as it does not clash with some other target isa name), s390_usable_hw or s390_hw_available is fine. Jakub
Re: [PATCH v4] Run tests only if the machine supports the instruction set.
On Tue, Dec 20, 2016 at 11:57:52AM -0800, Mike Stump wrote: > On Dec 20, 2016, at 6:10 AM, Dominik Vogtwrote: > > Right, it gets called even more often than one would think, and > > even with empty torture_current_options. The attached new patch > > (v3) removes -Ox options and superflous whitespace and caches that > > between calls if it's not empty. There's another, permanent cache > > for calls without any flags. With proper ordering of the torture > > options, the test program is built only a couple of times. > > Seems fine to me, but most other cases use the postfix _hw. Any > reason not use use _hw (and not _runable) on these? If not, > could you please use _hw instead. No specific reason other than lack of imagination. "s390_hw" is a bit too generic in my eyes -> the new names are: v4: * Renamed "s390_runnable" to "s390_useable_hw". * Renamed "z900_runnable" to "s390_z900_hw", Renamed "z10_runnable" to "s390_z10_hw", etc. Ciao Dominik ^_^ ^_^ -- Dominik Vogt IBM Germany gcc/ChangeLog-archlevel * config/s390/s390-c.c (s390_cpu_cpp_builtins_internal): Define __S390_ARCH_LEVEL__. gcc/testsuite/ChangeLog-setmem * gcc.target/s390/md/setmem_long-1.c: Use "s390_useable_hw". * gcc.target/s390/md/rXsbg_mode_sXl.c: Likewise. * gcc.target/s390/md/andc-splitter-1.c: Likewise. * gcc.target/s390/md/andc-splitter-2.c: Likewise. * lib/gcc-dg.exp (gcc-dg-runtest): Export torture_current_flags. * gcc.target/s390/s390.exp: Import torture_current_flags. (check_effective_target_s390_useable_hw): New. (check_effective_target_s390_z900_hw): New. (check_effective_target_s390_z990_hw): New. (check_effective_target_s390_z9_ec_hw): New. (check_effective_target_s390_z10_hw): New. (check_effective_target_s390_z196_hw): New. (check_effective_target_s390_zEC12_hw): New. (check_effective_target_s390_z13_hw): New. (check_effective_target_z10_instructions): Removed. (torture tests): Add optimization level without -march=. Reorder torture tests for good cache usage. >From 8ae8d6f48fbede4188a7e02c835fcf9d0b535da7 Mon Sep 17 00:00:00 2001 From: Dominik Vogt Date: Tue, 13 Dec 2016 10:21:08 +0100 Subject: [PATCH] S/390: Run md tests only if the machine supports the instruction set. --- gcc/config/s390/s390-c.c | 17 +++ gcc/testsuite/gcc.target/s390/md/andc-splitter-1.c | 19 +-- gcc/testsuite/gcc.target/s390/md/andc-splitter-2.c | 19 +-- gcc/testsuite/gcc.target/s390/md/rXsbg_mode_sXl.c | 4 +- gcc/testsuite/gcc.target/s390/md/setmem_long-1.c | 7 +- gcc/testsuite/gcc.target/s390/s390.exp | 170 ++--- gcc/testsuite/lib/gcc-dg.exp | 2 + 7 files changed, 198 insertions(+), 40 deletions(-) diff --git a/gcc/config/s390/s390-c.c b/gcc/config/s390/s390-c.c index fcf7477..e841365 100644 --- a/gcc/config/s390/s390-c.c +++ b/gcc/config/s390/s390-c.c @@ -320,6 +320,8 @@ s390_cpu_cpp_builtins_internal (cpp_reader *pfile, { s390_def_or_undef_macro (pfile, MASK_OPT_HTM, old_opts, opts, "__HTM__", "__HTM__"); + s390_def_or_undef_macro (pfile, MASK_OPT_VX, old_opts, opts, + "__S390_VX__", "__S390_VX__"); s390_def_or_undef_macro (pfile, MASK_ZVECTOR, old_opts, opts, "__VEC__=10301", "__VEC__"); s390_def_or_undef_macro (pfile, MASK_ZVECTOR, old_opts, opts, @@ -328,6 +330,21 @@ s390_cpu_cpp_builtins_internal (cpp_reader *pfile, s390_def_or_undef_macro (pfile, MASK_ZVECTOR, old_opts, opts, "__bool=__attribute__((s390_vector_bool)) unsigned", "__bool"); + { +char macro_def[64]; +int arch_level; +gcc_assert (s390_arch != PROCESSOR_NATIVE); +arch_level = (int)s390_arch + 3; +if (s390_arch >= PROCESSOR_2094_Z9_EC) + /* Z9_EC has the same level as Z9_109. */ + arch_level--; +/* Review when a new arch is added and increase the value. */ +char dummy[23 - 2 * PROCESSOR_max] __attribute__((unused)); +sprintf (macro_def, "__S390_ARCH_LEVEL__=%d", arch_level); +cpp_undef (pfile, "__S390_ARCH_LEVEL__"); +cpp_define (pfile, macro_def); + } + if (!flag_iso) { s390_def_or_undef_macro (pfile, MASK_ZVECTOR, old_opts, opts, diff --git a/gcc/testsuite/gcc.target/s390/md/andc-splitter-1.c b/gcc/testsuite/gcc.target/s390/md/andc-splitter-1.c index ed78921..3f0677c 100644 --- a/gcc/testsuite/gcc.target/s390/md/andc-splitter-1.c +++ b/gcc/testsuite/gcc.target/s390/md/andc-splitter-1.c @@ -1,7 +1,8 @@ /* Machine description pattern tests. */ -/* { dg-do run { target { lp64 } } } */ +/* { dg-do compile { target { lp64 } } } */ /* { dg-options "-mzarch -save-temps -dP" } */ +/* { dg-do run { target { lp64 && s390_useable_hw } } } */ /*
Re: [PATCH v3] Run tests only if the machine supports the instruction set.
On Dec 20, 2016, at 6:10 AM, Dominik Vogtwrote: > Right, it gets called even more often than one would think, and > even with empty torture_current_options. The attached new patch > (v3) removes -Ox options and superflous whitespace and caches that > between calls if it's not empty. There's another, permanent cache > for calls without any flags. With proper ordering of the torture > options, the test program is built only a couple of times. Seems fine to me, but most other cases use the postfix _hw. Any reason not use use _hw (and not _runable) on these? If not, could you please use _hw instead.
Re: [PATCH v3] Run tests only if the machine supports the instruction set.
On Tue, Dec 20, 2016 at 11:32:58AM +0100, Jakub Jelinek wrote: > On Tue, Dec 20, 2016 at 11:22:47AM +0100, Dominik Vogt wrote: > > On Mon, Dec 19, 2016 at 06:00:21PM +0100, Jakub Jelinek wrote: > > > On Mon, Dec 19, 2016 at 05:50:40PM +0100, Dominik Vogt wrote: > > > > * config/s390/s390-c.c (s390_cpu_cpp_builtins_internal): Define > > > > __S390_ARCH_LEVEL__. > > > > gcc/testsuite/ChangeLog-setmem > > > > > > > > * gcc.target/s390/md/setmem_long-1.c: Use "runnable". > > > > * gcc.target/s390/md/rXsbg_mode_sXl.c: Likewise. > > > > * gcc.target/s390/md/andc-splitter-1.c: Likewise. > > > > * gcc.target/s390/md/andc-splitter-2.c: Likewise. > > > > * lib/gcc-dg.exp (gcc-dg-runtest): Export torture_current_flags. > > > > * gcc.target/s390/s390.exp: Import torture_current_flags. > > > > (check_effective_target_runnable): New. > > > > > > Unless you want to add support for all targets in the runnable > > > effective target, I think it would be better to call it less generically, > > > s390_runnable or similar. Done. > > What do you think about the change in gcc-dg.exp? > > > > We couldn't > > decide whether it's a valid way of retrieving the flags needed for > > compiling s390_check_runnable or not. It would be nice to get all > > options that are relevant for the test case, including the ones > > from "dg-options" (etc.), but that probably requires larger > > changes to lib/*.exp. (The target specific check functions could > > be removed then.) > > I'm not a testsuite maintainer nor very good in tcl, so I think you want a > testsuite maintainer to ack it in any case. > But, I'd say you want something that will not be terribly expensive. > If you have an effective target that happens to get flags from the > current test, then that is necessarily non-cacheable, which would mean > in addition to compiling every test you also compile another proglet for it. > I think your current patch does that too, there is no caching, so it would > be desirable to cache the results; you should invalidate those caches when > torture_current_flags change. See e.g. et_cache uses in > target-supports.exp. You want to remember torture_current_flags for which > you've computed the s390_runnable et and if it changes, reset the cache. > The other *_runnable flags can be probably just normally cached (and you do, > by using check_runtime rather than check_runtime_nocache). Right, it gets called even more often than one would think, and even with empty torture_current_options. The attached new patch (v3) removes -Ox options and superflous whitespace and caches that between calls if it's not empty. There's another, permanent cache for calls without any flags. With proper ordering of the torture options, the test program is built only a couple of times. v3: * Cache test results. * Reorder torture tests for better caching. * Add ".machinemode zarch" to assembly file because the $flags are overridden by the board options. Ciao Dominik ^_^ ^_^ -- Dominik Vogt IBM Germany gcc/ChangeLog-archlevel * config/s390/s390-c.c (s390_cpu_cpp_builtins_internal): Define __S390_ARCH_LEVEL__. gcc/testsuite/ChangeLog-setmem * gcc.target/s390/md/setmem_long-1.c: Use "s390_runnable". * gcc.target/s390/md/rXsbg_mode_sXl.c: Likewise. * gcc.target/s390/md/andc-splitter-1.c: Likewise. * gcc.target/s390/md/andc-splitter-2.c: Likewise. * lib/gcc-dg.exp (gcc-dg-runtest): Export torture_current_flags. * gcc.target/s390/s390.exp: Import torture_current_flags. (check_effective_target_s390_runnable): New. (check_effective_target_z900_runnable): New. (check_effective_target_z990_runnable): New. (check_effective_target_z9_ec_runnable): New. (check_effective_target_z10_runnable): New. (check_effective_target_z196_runnable): New. (check_effective_target_zEC12_runnable): New. (check_effective_target_z13_runnable): New. (check_effective_target_z10_instructions): Removed. (MD_TEST_OPTS): Add optimization level without -march=. >From 4bd55b91b0487590bc9c9bde60664e5d94d2dc08 Mon Sep 17 00:00:00 2001 From: Dominik VogtDate: Tue, 13 Dec 2016 10:21:08 +0100 Subject: [PATCH] S/390: Run md tests only if the machine supports the instruction set. --- gcc/config/s390/s390-c.c | 17 +++ gcc/testsuite/gcc.target/s390/md/andc-splitter-1.c | 19 +-- gcc/testsuite/gcc.target/s390/md/andc-splitter-2.c | 19 +-- gcc/testsuite/gcc.target/s390/md/rXsbg_mode_sXl.c | 4 +- gcc/testsuite/gcc.target/s390/md/setmem_long-1.c | 7 +- gcc/testsuite/gcc.target/s390/s390.exp | 170 ++--- gcc/testsuite/lib/gcc-dg.exp | 2 + 7 files changed, 198 insertions(+), 40 deletions(-) diff --git a/gcc/config/s390/s390-c.c b/gcc/config/s390/s390-c.c index
Re: [PATCH v2] Run tests only if the machine supports the instruction set.
On Tue, Dec 20, 2016 at 11:22:47AM +0100, Dominik Vogt wrote: > On Mon, Dec 19, 2016 at 06:00:21PM +0100, Jakub Jelinek wrote: > > On Mon, Dec 19, 2016 at 05:50:40PM +0100, Dominik Vogt wrote: > > > * config/s390/s390-c.c (s390_cpu_cpp_builtins_internal): Define > > > __S390_ARCH_LEVEL__. > > > gcc/testsuite/ChangeLog-setmem > > > > > > * gcc.target/s390/md/setmem_long-1.c: Use "runnable". > > > * gcc.target/s390/md/rXsbg_mode_sXl.c: Likewise. > > > * gcc.target/s390/md/andc-splitter-1.c: Likewise. > > > * gcc.target/s390/md/andc-splitter-2.c: Likewise. > > > * lib/gcc-dg.exp (gcc-dg-runtest): Export torture_current_flags. > > > * gcc.target/s390/s390.exp: Import torture_current_flags. > > > (check_effective_target_runnable): New. > > > > Unless you want to add support for all targets in the runnable > > effective target, I think it would be better to call it less generically, > > s390_runnable or similar. > > Fair enough. > > What do you think about the change in gcc-dg.exp? > > We couldn't > decide whether it's a valid way of retrieving the flags needed for > compiling s390_check_runnable or not. It would be nice to get all > options that are relevant for the test case, including the ones > from "dg-options" (etc.), but that probably requires larger > changes to lib/*.exp. (The target specific check functions could > be removed then.) I'm not a testsuite maintainer nor very good in tcl, so I think you want a testsuite maintainer to ack it in any case. But, I'd say you want something that will not be terribly expensive. If you have an effective target that happens to get flags from the current test, then that is necessarily non-cacheable, which would mean in addition to compiling every test you also compile another proglet for it. I think your current patch does that too, there is no caching, so it would be desirable to cache the results; you should invalidate those caches when torture_current_flags change. See e.g. et_cache uses in target-supports.exp. You want to remember torture_current_flags for which you've computed the s390_runnable et and if it changes, reset the cache. The other *_runnable flags can be probably just normally cached (and you do, by using check_runtime rather than check_runtime_nocache). Jakub
Re: [PATCH v2] Run tests only if the machine supports the instruction set.
On Mon, Dec 19, 2016 at 06:00:21PM +0100, Jakub Jelinek wrote: > On Mon, Dec 19, 2016 at 05:50:40PM +0100, Dominik Vogt wrote: > > * config/s390/s390-c.c (s390_cpu_cpp_builtins_internal): Define > > __S390_ARCH_LEVEL__. > > gcc/testsuite/ChangeLog-setmem > > > > * gcc.target/s390/md/setmem_long-1.c: Use "runnable". > > * gcc.target/s390/md/rXsbg_mode_sXl.c: Likewise. > > * gcc.target/s390/md/andc-splitter-1.c: Likewise. > > * gcc.target/s390/md/andc-splitter-2.c: Likewise. > > * lib/gcc-dg.exp (gcc-dg-runtest): Export torture_current_flags. > > * gcc.target/s390/s390.exp: Import torture_current_flags. > > (check_effective_target_runnable): New. > > Unless you want to add support for all targets in the runnable > effective target, I think it would be better to call it less generically, > s390_runnable or similar. Fair enough. What do you think about the change in gcc-dg.exp? We couldn't decide whether it's a valid way of retrieving the flags needed for compiling s390_check_runnable or not. It would be nice to get all options that are relevant for the test case, including the ones from "dg-options" (etc.), but that probably requires larger changes to lib/*.exp. (The target specific check functions could be removed then.) Ciao Dominik ^_^ ^_^ -- Dominik Vogt IBM Germany
Re: [PATCH v2] Run tests only if the machine supports the instruction set.
On Mon, Dec 19, 2016 at 05:50:40PM +0100, Dominik Vogt wrote: > * config/s390/s390-c.c (s390_cpu_cpp_builtins_internal): Define > __S390_ARCH_LEVEL__. > gcc/testsuite/ChangeLog-setmem > > * gcc.target/s390/md/setmem_long-1.c: Use "runnable". > * gcc.target/s390/md/rXsbg_mode_sXl.c: Likewise. > * gcc.target/s390/md/andc-splitter-1.c: Likewise. > * gcc.target/s390/md/andc-splitter-2.c: Likewise. > * lib/gcc-dg.exp (gcc-dg-runtest): Export torture_current_flags. > * gcc.target/s390/s390.exp: Import torture_current_flags. > (check_effective_target_runnable): New. Unless you want to add support for all targets in the runnable effective target, I think it would be better to call it less generically, s390_runnable or similar. Jakub
Re: [PATCH v2] Run tests only if the machine supports the instruction set.
On Mon, Dec 19, 2016 at 03:28:06PM +0100, Dominik Vogt wrote: > The attached patch is specific to S/390 but contains a small > common code change in gcc-dg.exp. It fixes the notorious problem > of md tests running on an S/390 machine that does not support the > z13 instruction set. > > Bootstrapped and tested on s390x biarch. Version 2 with results of internal discussion: * Renamed __s390_arh_level__ to upper case. * Replaced __VECTOR__ with new macro __S390_VX__. * Added individual (but currently unused) effective-target functions for the various architectures. Ciao Dominik ^_^ ^_^ -- Dominik Vogt IBM Germany gcc/ChangeLog-archlevel * config/s390/s390-c.c (s390_cpu_cpp_builtins_internal): Define __S390_ARCH_LEVEL__. gcc/testsuite/ChangeLog-setmem * gcc.target/s390/md/setmem_long-1.c: Use "runnable". * gcc.target/s390/md/rXsbg_mode_sXl.c: Likewise. * gcc.target/s390/md/andc-splitter-1.c: Likewise. * gcc.target/s390/md/andc-splitter-2.c: Likewise. * lib/gcc-dg.exp (gcc-dg-runtest): Export torture_current_flags. * gcc.target/s390/s390.exp: Import torture_current_flags. (check_effective_target_runnable): New. (check_effective_target_z900_runnable): New. (check_effective_target_z990_runnable): New. (check_effective_target_z9_ec_runnable): New. (check_effective_target_z10_runnable): New. (check_effective_target_z196_runnable): New. (check_effective_target_zEC12_runnable): New. (check_effective_target_z13_runnable): New. (check_effective_target_z10_instructions): Removed. (MD_TEST_OPTS): Add optimization level without -march=. >From 7c74ce8212fa85272e1928b2d2df44b74cba5d0a Mon Sep 17 00:00:00 2001 From: Dominik VogtDate: Tue, 13 Dec 2016 10:21:08 +0100 Subject: [PATCH] S/390: Run md tests only if the machine supports the instruction set. --- gcc/config/s390/s390-c.c | 17 +++ gcc/testsuite/gcc.target/s390/md/andc-splitter-1.c | 19 ++-- gcc/testsuite/gcc.target/s390/md/andc-splitter-2.c | 19 ++-- gcc/testsuite/gcc.target/s390/md/rXsbg_mode_sXl.c | 4 +- gcc/testsuite/gcc.target/s390/md/setmem_long-1.c | 7 +- gcc/testsuite/gcc.target/s390/s390.exp | 126 ++--- gcc/testsuite/lib/gcc-dg.exp | 2 + 7 files changed, 155 insertions(+), 39 deletions(-) diff --git a/gcc/config/s390/s390-c.c b/gcc/config/s390/s390-c.c index fcf7477..e841365 100644 --- a/gcc/config/s390/s390-c.c +++ b/gcc/config/s390/s390-c.c @@ -320,6 +320,8 @@ s390_cpu_cpp_builtins_internal (cpp_reader *pfile, { s390_def_or_undef_macro (pfile, MASK_OPT_HTM, old_opts, opts, "__HTM__", "__HTM__"); + s390_def_or_undef_macro (pfile, MASK_OPT_VX, old_opts, opts, + "__S390_VX__", "__S390_VX__"); s390_def_or_undef_macro (pfile, MASK_ZVECTOR, old_opts, opts, "__VEC__=10301", "__VEC__"); s390_def_or_undef_macro (pfile, MASK_ZVECTOR, old_opts, opts, @@ -328,6 +330,21 @@ s390_cpu_cpp_builtins_internal (cpp_reader *pfile, s390_def_or_undef_macro (pfile, MASK_ZVECTOR, old_opts, opts, "__bool=__attribute__((s390_vector_bool)) unsigned", "__bool"); + { +char macro_def[64]; +int arch_level; +gcc_assert (s390_arch != PROCESSOR_NATIVE); +arch_level = (int)s390_arch + 3; +if (s390_arch >= PROCESSOR_2094_Z9_EC) + /* Z9_EC has the same level as Z9_109. */ + arch_level--; +/* Review when a new arch is added and increase the value. */ +char dummy[23 - 2 * PROCESSOR_max] __attribute__((unused)); +sprintf (macro_def, "__S390_ARCH_LEVEL__=%d", arch_level); +cpp_undef (pfile, "__S390_ARCH_LEVEL__"); +cpp_define (pfile, macro_def); + } + if (!flag_iso) { s390_def_or_undef_macro (pfile, MASK_ZVECTOR, old_opts, opts, diff --git a/gcc/testsuite/gcc.target/s390/md/andc-splitter-1.c b/gcc/testsuite/gcc.target/s390/md/andc-splitter-1.c index ed78921..9c41ac4 100644 --- a/gcc/testsuite/gcc.target/s390/md/andc-splitter-1.c +++ b/gcc/testsuite/gcc.target/s390/md/andc-splitter-1.c @@ -1,7 +1,8 @@ /* Machine description pattern tests. */ -/* { dg-do run { target { lp64 } } } */ +/* { dg-do compile { target { lp64 } } } */ /* { dg-options "-mzarch -save-temps -dP" } */ +/* { dg-do run { target { lp64 && runnable } } } */ /* Skip test if -O0 is present on the command line: { dg-skip-if "" { *-*-* } { "-O0" } { "" } } @@ -13,26 +14,26 @@ __attribute__ ((noinline)) unsigned long andc_vv(unsigned long a, unsigned long b) { return ~b & a; } -/* { dg-final { scan-assembler ":15 .\* \{\\*anddi3\}" } } */ -/* { dg-final { scan-assembler ":15 .\* \{\\*xordi3\}" } } */ +/* { dg-final { scan-assembler ":16 .\* \{\\*anddi3\}" } } */ +/* { dg-final { scan-assembler ":16 .\* \{\\*xordi3\}" } } */
Run tests only if the machine supports the instruction set.
The attached patch is specific to S/390 but contains a small common code change in gcc-dg.exp. It fixes the notorious problem of md tests running on an S/390 machine that does not support the z13 instruction set. Bootstrapped and tested on s390x biarch. Ciao Dominik ^_^ ^_^ -- Dominik Vogt IBM Germany gcc/ChangeLog-archlevel * config/s390/s390-c.c (s390_cpu_cpp_builtins_internal): Define __s390_arch_level__. gcc/testsuite/ChangeLog-setmem * gcc.target/s390/md/setmem_long-1.c: Use "runnable". * gcc.target/s390/md/rXsbg_mode_sXl.c: Likewise. * gcc.target/s390/md/andc-splitter-1.c: Likewise. * gcc.target/s390/md/andc-splitter-2.c: Likewise. * lib/gcc-dg.exp (gcc-dg-runtest): Export torture_current_flags. * gcc.target/s390/s390.exp: Import torture_current_flags. (check_effective_target_runnable): New. (check_effective_target_z10_instructions): Removed. (MD_TEST_OPTS): Add optimization level without -march=. >From 21e99fb09c5e8350892d99e6c351515333594aae Mon Sep 17 00:00:00 2001 From: Dominik VogtDate: Tue, 13 Dec 2016 10:21:08 +0100 Subject: [PATCH] S/390: Run md tests only if the machine supports the instruction set. --- gcc/config/s390/s390-c.c | 15 ++ gcc/testsuite/gcc.target/s390/md/andc-splitter-1.c | 19 +++ gcc/testsuite/gcc.target/s390/md/andc-splitter-2.c | 19 +++ gcc/testsuite/gcc.target/s390/md/rXsbg_mode_sXl.c | 4 +- gcc/testsuite/gcc.target/s390/md/setmem_long-1.c | 7 +-- gcc/testsuite/gcc.target/s390/s390.exp | 60 -- gcc/testsuite/lib/gcc-dg.exp | 2 + 7 files changed, 87 insertions(+), 39 deletions(-) diff --git a/gcc/config/s390/s390-c.c b/gcc/config/s390/s390-c.c index fcf7477..8a9ea79 100644 --- a/gcc/config/s390/s390-c.c +++ b/gcc/config/s390/s390-c.c @@ -328,6 +328,21 @@ s390_cpu_cpp_builtins_internal (cpp_reader *pfile, s390_def_or_undef_macro (pfile, MASK_ZVECTOR, old_opts, opts, "__bool=__attribute__((s390_vector_bool)) unsigned", "__bool"); + { +char macro_def[64]; +int arch_level; +gcc_assert (s390_arch != PROCESSOR_NATIVE); +arch_level = (int)s390_arch + 3; +if (s390_arch >= PROCESSOR_2094_Z9_EC) + /* Z9_EC has the same level as Z9_109. */ + arch_level--; +/* Review when a new arch is added and increase the value. */ +char dummy[23 - 2 * PROCESSOR_max] __attribute__((unused)); +sprintf (macro_def, "__s390_arch_level__=%d", arch_level); +cpp_undef (pfile, "__s390_arch_level__"); +cpp_define (pfile, macro_def); + } + if (!flag_iso) { s390_def_or_undef_macro (pfile, MASK_ZVECTOR, old_opts, opts, diff --git a/gcc/testsuite/gcc.target/s390/md/andc-splitter-1.c b/gcc/testsuite/gcc.target/s390/md/andc-splitter-1.c index ed78921..9c41ac4 100644 --- a/gcc/testsuite/gcc.target/s390/md/andc-splitter-1.c +++ b/gcc/testsuite/gcc.target/s390/md/andc-splitter-1.c @@ -1,7 +1,8 @@ /* Machine description pattern tests. */ -/* { dg-do run { target { lp64 } } } */ +/* { dg-do compile { target { lp64 } } } */ /* { dg-options "-mzarch -save-temps -dP" } */ +/* { dg-do run { target { lp64 && runnable } } } */ /* Skip test if -O0 is present on the command line: { dg-skip-if "" { *-*-* } { "-O0" } { "" } } @@ -13,26 +14,26 @@ __attribute__ ((noinline)) unsigned long andc_vv(unsigned long a, unsigned long b) { return ~b & a; } -/* { dg-final { scan-assembler ":15 .\* \{\\*anddi3\}" } } */ -/* { dg-final { scan-assembler ":15 .\* \{\\*xordi3\}" } } */ +/* { dg-final { scan-assembler ":16 .\* \{\\*anddi3\}" } } */ +/* { dg-final { scan-assembler ":16 .\* \{\\*xordi3\}" } } */ __attribute__ ((noinline)) unsigned long andc_pv(unsigned long *a, unsigned long b) { return ~b & *a; } -/* { dg-final { scan-assembler ":21 .\* \{\\*anddi3\}" } } */ -/* { dg-final { scan-assembler ":21 .\* \{\\*xordi3\}" } } */ +/* { dg-final { scan-assembler ":22 .\* \{\\*anddi3\}" } } */ +/* { dg-final { scan-assembler ":22 .\* \{\\*xordi3\}" } } */ __attribute__ ((noinline)) unsigned long andc_vp(unsigned long a, unsigned long *b) { return ~*b & a; } -/* { dg-final { scan-assembler ":27 .\* \{\\*anddi3\}" } } */ -/* { dg-final { scan-assembler ":27 .\* \{\\*xordi3\}" } } */ +/* { dg-final { scan-assembler ":28 .\* \{\\*anddi3\}" } } */ +/* { dg-final { scan-assembler ":28 .\* \{\\*xordi3\}" } } */ __attribute__ ((noinline)) unsigned long andc_pp(unsigned long *a, unsigned long *b) { return ~*b & *a; } -/* { dg-final { scan-assembler ":33 .\* \{\\*anddi3\}" } } */ -/* { dg-final { scan-assembler ":33 .\* \{\\*xordi3\}" } } */ +/* { dg-final { scan-assembler ":34 .\* \{\\*anddi3\}" } } */ +/* { dg-final { scan-assembler ":34 .\* \{\\*xordi3\}" } } */ /* { dg-final { scan-assembler-times "\tngr\?k\?\t" 4 } } */ /* { dg-final { scan-assembler-times