This is the mail archive of the gcc-patches@gcc.gnu.org mailing list for the GCC project.
Index Nav: | [Date Index] [Subject Index] [Author Index] [Thread Index] | |
---|---|---|
Message Nav: | [Date Prev] [Date Next] | [Thread Prev] [Thread Next] |
Other format: | [Raw text] |
On Mon, Jul 24, 2017 at 05:21:15AM -0500, Segher Boessenkool wrote: > Hi Mike, > > On Sat, Jul 22, 2017 at 02:46:04AM -0400, Michael Meissner wrote: > > One thing that I did was continue to define __UPPER_REGS_{DF,SF,DI}__ that were > > previously defined. I can delete them if desired and perhaps poison the names > > so that any use if flaged. > > I don't think anything uses it. Certainly nothing *should* use it: it > isn't documented anywhere :-) (Google doesn't find anything either, fwiw). > > You can delete them I think. Poisoning is overkill. Ok, deleted. > > Future patches will include removal of the TARGET_UPPER_REGS_* macros in the > > various files. I also plan to remove -mvsx-small-integer, and the various > > -mpower9-dform* options. > > Yay! Thank you. > > > Is this ok for trunk? At present, I do not intend to back port this to GCC 7 > > (but if the maintainers want that, I can do it). > > I don't think that would help much, so yeah, let's not. > > > @@ -58,7 +56,6 @@ > > | OPTION_MASK_HTM \ > > | OPTION_MASK_QUAD_MEMORY \ > > | OPTION_MASK_QUAD_MEMORY_ATOMIC \ > > - | OPTION_MASK_UPPER_REGS_SF \ > > | OPTION_MASK_VSX_SMALL_INTEGER) > > > > /* Add ISEL back into ISA 3.0, since it is supposed to be a win. Do not add > > The OPTION_MASK_QUAD_MEMORY_ATOMIC line has errant leading spaces, maybe > you could fix that while at it. Fixed. > > + /* Note, previously __UPPER_REGS_DF__ was defined if the option > > + -mupper-regs-df was used and it was on by default for -mvsx. That option > > + is now eliminated, so set __UPPER_REGS_DF__ based on whether VSX was > > + set. */ > > "That option" reads as refering to -mvsx. > > > --- gcc/config/rs6000/rs6000.c (.../svn+ssh://meissner@gcc.gnu.org/svn/gcc/trunk/gcc/config/rs6000) (revision 250405) > > +++ gcc/config/rs6000/rs6000.c (.../gcc/config/rs6000) (working copy) > > @@ -2908,8 +2908,7 @@ rs6000_setup_reg_addr_masks (void) > > && !FLOAT128_VECTOR_P (m2) > > && !complex_p > > && !small_int_vsx_p > > - && (m2 != DFmode || !TARGET_UPPER_REGS_DF) > > - && (m2 != SFmode || !TARGET_UPPER_REGS_SF)) > > + && (m2 != DFmode || !TARGET_UPPER_REGS_DF)) > > { > > addr_mask |= RELOAD_REG_PRE_INCDEC; > > > Why are you leaving DF here? (A later patch will take care of it, but > is it more than an oversight?) While this likely would be removed in the next patch, I deleted it now for consistancy. > > +/* { dg-final { scan-assembler-times {\mfctidz\M|\mxscvdpsxds\M} 2 } } */ > > You could write {\m(fctidz|xscvdpsxds)\M} which may be easier to read. > Come to think of it, we could do e.g. {\m(lwz)\M} as well. Not sure > which is nicer. > > Looks good, please commit! Thanks again, I wasn't sure which form of regex (shell, grep, egrep) was used, and what levels of quoting was needed, so I wrote it the other way. Here is the patch I committed: [gcc] 2017-07-24 Michael Meissner <meissner@linux.vnet.ibm.com> * config/rs6000/rs6000-cpus.def (ISA_2_6_MASKS_SERVER): Delete upper-regs options. (ISA_2_7_MASKS_SERVER): Likewise. (ISA_3_0_MASKS_IEEE): Likewise. (OTHER_P8_VECTOR_MASKS): Likewise. (OTHER_VSX_VECTOR_MASKS): Likewise. (POWERPC_MASKS): Likewise. (power7 cpu): Use ISA_2_6_MASKS_SERVER instead of using a duplicate list of options. * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Remove explicit -mupper-regs options. * config/rs6000/rs6000.opt (-mvsx-scalar-memory): Delete -mupper-regs* options. Delete -mvsx-scalar-memory, which was an alias for -mupper-regs-df. * config/rs6000/rs6000.c (rs6000_setup_reg_addr_masks): Likewise. (rs6000_init_hard_regno_mode_ok): Likewise. (rs6000_option_override_internal): Likewise. (rs6000_opt_masks): Likewise. * config/rs6000/rs6000.h (TARGET_UPPER_REGS_DF): Define upper regs options in terms of whether -mvsx or -mpower8-vector was used. (TARGET_UPPER_REGS_DI): Likewise. (TARGET_UPPER_REGS_SF): Likewise. * doc/invoke.texi (RS/6000 and PowerPC Options): Delete the -mupper-regs-* options. [gcc/testsuite] 2017-07-24 Michael Meissner <meissner@linux.vnet.ibm.com> * gcc.target/powerpc/pr65849-1.c: Delete, test no longer valid since the upper-regs options have been deleted. * gcc.target/powerpc/pr65849-2.c: Likewise. * gcc.target/powerpc/pr80099-1.c: Likewise. * gcc.target/powerpc/pr80099-2.c: Likewise. * gcc.target/powerpc/pr80099-3.c: Likewise. * gcc.target/powerpc/pr80099-4.c: Likewise. * gcc.target/powerpc/pr80099-5.c: Likewise. * gcc.target/powerpc/builtins-2-p9-runnable.c: Update test to support removal of the upper-regs options. * gcc.target/powerpc/p8vector-fp.c: Likewise. * gcc.target/powerpc/p8vector-ldst.c: Likewise. * gcc.target/powerpc/p9-dimode1.c: Likewise. * gcc.target/powerpc/p9-dimode2.c: Likewise. * gcc.target/powerpc/ppc-fpconv-1.c: Likewise. * gcc.target/powerpc/ppc-fpconv-10.c: Likewise. * gcc.target/powerpc/ppc-fpconv-5.c: Likewise. * gcc.target/powerpc/ppc-fpconv-9.c: Likewise. * gcc.target/powerpc/ppc-round.c: Likewise. * gcc.target/powerpc/pr71720.c: Likewise. * gcc.target/powerpc/pr72853.c: Likewise. * gcc.target/powerpc/pr79907.c: Likewise. * gcc.target/powerpc/pr78953.c: Likewise. * gcc.target/powerpc/upper-regs-df.c: Likewise. * gcc.target/powerpc/upper-regs-sf.c: Likewise. * gcc.target/powerpc/vec-extract-1.c: Likewise. * gcc.target/powerpc/vec-init-3.c: Likewise. * gcc.target/powerpc/vec-init-6.c: Likewise. * gcc.target/powerpc/vec-init-7.c: Likewise. * gcc.target/powerpc/vec-set-char.c: Likewise. * gcc.target/powerpc/vec-set-int.c: Likewise. * gcc.target/powerpc/vec-set-short.c: Likewise. Now to remove the remaining TARGET_UPPER_REGS_* references. -- Michael Meissner, IBM IBM, M/S 2506R, 550 King Street, Littleton, MA 01460-6245, USA email: meissner@linux.vnet.ibm.com, phone: +1 (978) 899-4797
Attachment:
cleanup.patch002b
Description: Text document
Index Nav: | [Date Index] [Subject Index] [Author Index] [Thread Index] | |
---|---|---|
Message Nav: | [Date Prev] [Date Next] | [Thread Prev] [Thread Next] |