or1k: Add mcmodel option to handle large GOTs
PATCH 1/1 or1k: Add mcmodel option to handle large GOTs
When building libgeos we get an error with:
linux-uclibc/9.3.0/crtbeginS.o: in function `__do_global_dtors_aux':
crtstuff.c:(.text+0x118): relocation truncated to fit: R_OR1K_GOT16 against symbol `__cxa_finalize' defined in .text section in
/home/shorne/work/openrisc/3eb9f9d0f6d8274b2d19753c006bd83f7d536e3c/output/host/or1k-buildroot-linux-uclibc/sysroot/lib/libc.so.This is caused by GOT code having a limit of 64k. In OpenRISC this looks to be the only relocation code pattern to be limited to 64k.
This patch allows specifying a new option -mcmodel=large which can be used to generate 2 more instructions to construct 32-bit addresses for up to 4G GOTs.
gcc/ChangeLog:
PR 99783
* config/or1k/or1k-opts.h: New file.
* config/or1k/or1k.c (or1k_legitimize_address_1, print_reloc):
Support generating gotha relocations if -mcmodel=large is specified.
* config/or1k/or1k.h (TARGET_CMODEL_SMALL, TARGET_CMODEL_LARGE):
New macros.
* config/or1k/or1k.opt (mcmodel=): New option.
* doc/invoke.text (OpenRISC Options): Document mcmodel.
---
This depends on the binutils-gdb patch sent here:
- https://sourceware.org/pipermail/binutils/2021-April/116155.html
gcc/config/or1k/or1k-opts.h | 30 ++++++++++++++++++++++++++++++ gcc/config/or1k/or1k.c | 11 +++++++++-- gcc/config/or1k/or1k.h | 7 +++++++ gcc/config/or1k/or1k.opt | 19 +++++++++++++++++++ gcc/doc/invoke.texi | 12 +++++++++++- 5 files changed, 76 insertions(+), 3 deletions(-) create mode 100644 gcc/config/or1k/or1k-opts.h
gcc/config/or1k/or1k-opts.h new file +30 -0
@@ -0,0 +1,30 @@ +/* Definitions for option handling for OpenRISC. +Copyright (C) 2021 Free Software Foundation, Inc. +Contributed by Stafford Horne. + +This file is part of GCC. + +GCC is free software; you can redistribute it and/or modify it +under the terms of the GNU General Public License as published +by the Free Software Foundation; either version 3, or (at your +option) any later version. + +GCC is distributed in the hope that it will be useful, but WITHOUT +ANY WARRANTY; without even the implied warranty of MERCHANTABILITY +or FITNESS FOR A PARTICULAR PURPOSE. See the GNU General Public +License for more details. + +You should have received a copy of the GNU General Public License +along with GCC; see the file COPYING3. If not see +<http://www.gnu.org/licenses/>. */ + +#ifndef GCC_OR1K_OPTS_H +#define GCC_OR1K_OPTS_H + +/* The OpenRISC code generation models available. */ +enum or1k_cmodel_type { +CMODEL_SMALL , +CMODEL_LARGE +}; + +#endif /* GCC_OR1K_OPTS_H */
gcc/config/or1k/or1k.c +9 -2
@@ -750,7 +750,14 @@ or1k_legitimize_address_1 (rtx x, rtx scratch) { base =gen_sym_unspec (base,UNSPEC_GOT ); crtl->uses_pic_offset_table =1 ; - t2 =gen_rtx_LO_SUM (Pmode, pic_offset_table_rtx, base); +if (TARGET_CMODEL_LARGE ) + { +emit_insn (gen_rtx_SET (t1,gen_rtx_HIGH (Pmode, base))); +emit_insn (gen_add3_insn (t1, t1, pic_offset_table_rtx)); + t2 =gen_rtx_LO_SUM (Pmode, t1, base); + } +else + t2 =gen_rtx_LO_SUM (Pmode, pic_offset_table_rtx, base); t2 =gen_const_mem (Pmode, t2);emit_insn (gen_rtx_SET (t1, t2)); base = t1;
@@ -1089,7 +1096,7 @@ print_reloc (FILE *stream, rtx x, HOST_WIDE_INT add, reloc_kind kind)no special markup. */ static const char *const relocs[RKIND_MAX ][RTYPE_MAX ] = { {"lo" ,"got" ,"gotofflo" ,"tpofflo" ,"gottpofflo" ,"tlsgdlo" }, - {"ha" ,NULL ,"gotoffha" ,"tpoffha" ,"gottpoffha" ,"tlsgdhi" }, + {"ha" ,"gotha" ,"gotoffha" ,"tpoffha" ,"gottpoffha" ,"tlsgdhi" }, };reloc_type type =RTYPE_DIRECT ;
gcc/config/or1k/or1k.h +7 -0
@@ -21,6 +21,8 @@#ifndef GCC_OR1K_H #define GCC_OR1K_H +#include "config/or1k/or1k-opts.h" +/* Names to predefine in the preprocessor for this target machine. */ #define TARGET_CPU_CPP_BUILTINS () \ do \
@@ -37,6 +39,11 @@ } \ while (0) +#define TARGET_CMODEL_SMALL \ + (or1k_code_model == CMODEL_SMALL) +#define TARGET_CMODEL_LARGE \ + (or1k_code_model == CMODEL_LARGE) +/* Storage layout. */ #define DEFAULT_SIGNED_CHAR 1
gcc/config/or1k/or1k.opt +19 -0
@@ -21,6 +21,9 @@ ; See the GCC internals manual (options.texi) for a description of ; this file's format. +HeaderInclude +config/or1k/or1k-opts.h + mhard-div Target RejectNegative InverseMask(SOFT_DIV) Enable generation of hardware divide (l.div, l.divu) instructions. This is the
@@ -63,6 +66,22 @@ When -mhard-float is selected, enables generation of unordered floating point compare and set flag (lf.sfun*) instructions. By default functions from libgcc are used to perform unordered floating point compare and set flag operations. +mcmodel= +Target RejectNegative Joined Enum(or1k_cmodel_type) Var(or1k_code_model) Init(CMODEL_SMALL) +Specify the code model used for accessing memory addresses. Specifying large +enables generating binaries with large global offset tables. By default the +value is small. + +Enum +Name(or1k_cmodel_type) Type(enum or1k_cmodel_type) +Known code model types (for use with the -mcmodel= option): + +EnumValue +Enum(or1k_cmodel_type) String(small) Value(CMODEL_SMALL) + +EnumValue +Enum(or1k_cmodel_type) String(large) Value(CMODEL_LARGE) + mcmov Target RejectNegative Mask(CMOV) Enable generation of conditional move (l.cmov) instructions. By default the
gcc/doc/invoke.texi +11 -1
@@ -1135,7 +1135,8 @@ Objective-C and Objective-C++ Dialects}. @gccoptlist{-mboard=@var{name} -mnewlib -mhard-mul -mhard-div @gol -msoft-mul -msoft-div @gol -msoft-float -mhard-float -mdouble-float -munordered-float @gol --mcmov -mror -mrori -msext -msfimm -mshftimm} +-mcmov -mror -mrori -msext -msfimm -mshftimm @gol +-mcmodel=@var{code-model}} @emph{PDP-11 Options} @gccoptlist{-mfpu -msoft-float -mac0 -mno-ac0 -m40 -m45 -m10 @gol
@@ -26406,6 +26407,15 @@ Enable generation of shift with immediate (@code{l.srai}, @code{l.srli}, @code{l.slli}) instructions. By default extra instructions will be generated to store the immediate to a register first. + at item -mcmodel=small + at opindex mcmodel=small +Generate OpenRISC code for the small model: The GOT is limited to 64k. This is +the default model. + + at item -mcmodel=large + at opindex mcmodel=large +Generate OpenRISC code for the large model: The GOT may grow up to 4G in size. + @end table
Signature · 1 line
2.26.2
PATCH 1/1 or1k: Add mcmodel option to handle large GOTs
Quoted 15 lines · Stafford Horne via Gcc-patches
When building libgeos we get an error with:
linux-uclibc/9.3.0/crtbeginS.o: in function `__do_global_dtors_aux': crtstuff.c:(.text+0x118): relocation truncated to fit: R_OR1K_GOT16 against symbol `__cxa_finalize' defined in .text section in /home/shorne/work/openrisc/3eb9f9d0f6d8274b2d19753c006bd83f7d536e3c/output/host/or1k-buildroot-linux-uclibc/sysroot/lib/libc.so.This is caused by GOT code having a limit of 64k. In OpenRISC this looks to be the only relocation code pattern to be limited to 64k.
This patch allows specifying a new option -mcmodel=large which can be used to generate 2 more instructions to construct 32-bit addresses for up to 4G GOTs.
gcc/ChangeLog:
Quoted 9 lines
PR 99783
*config/or1k/or1k-opts.h: New file.
*config/or1k/or1k.c(or1k_legitimize_address_1,print_reloc):
Support generating gotha relocations if -mcmodel=large is specified.
*config/or1k/or1k.h(TARGET_CMODEL_SMALL,TARGET_CMODEL_LARGE):
New macros.
* config/or1k/or1k.opt (mcmodel=): New option.
* doc/invoke.text (OpenRISC Options): Document mcmodel.
Note you're the port maintainer, so you don't need approval to commit this to GCC.
Jeff
PATCH 1/1 or1k: Add mcmodel option to handle large GOTs
Quoted 27 lines · Jeff Law
Quoted 15 lines · Stafford Horne via Gcc-patches
When building libgeos we get an error with:
linux-uclibc/9.3.0/crtbeginS.o: in function `__do_global_dtors_aux': crtstuff.c:(.text+0x118): relocation truncated to fit: R_OR1K_GOT16 against symbol `__cxa_finalize' defined in .text section in /home/shorne/work/openrisc/3eb9f9d0f6d8274b2d19753c006bd83f7d536e3c/output/host/or1k-buildroot-linux-uclibc/sysroot/lib/libc.so.This is caused by GOT code having a limit of 64k. In OpenRISC this looks to be the only relocation code pattern to be limited to 64k.
This patch allows specifying a new option -mcmodel=large which can be used to generate 2 more instructions to construct 32-bit addresses for up to 4G GOTs.
gcc/ChangeLog:
Quoted 9 lines
PR 99783
*config/or1k/or1k-opts.h: New file.
*config/or1k/or1k.c(or1k_legitimize_address_1,print_reloc):
Support generating gotha relocations if -mcmodel=large is specified.
*config/or1k/or1k.h(TARGET_CMODEL_SMALL,TARGET_CMODEL_LARGE):
New macros.
* config/or1k/or1k.opt (mcmodel=): New option.
* doc/invoke.text (OpenRISC Options): Document mcmodel.
Quoted 2 lines
Note you're the port maintainer, so you don't need approval to commit this to GCC.
Thanks for confirming, I know that after Segher told me with some previous patches. :)
For this series Giulio picked it up and tested it. We found some issues and I am going to fix with a V2. I should commit that version, after the binutils changes are in.
-Stafford
PATCH 1/1 or1k: Add mcmodel option to handle large GOTs
Quoted 26 lines · Stafford Horne
Quoted 25 lines · Jeff Law
Quoted 24 lines · Stafford Horne via Gcc-patches
When building libgeos we get an error with:
linux-uclibc/9.3.0/crtbeginS.o: in function `__do_global_dtors_aux': crtstuff.c:(.text+0x118): relocation truncated to fit: R_OR1K_GOT16 against symbol `__cxa_finalize' defined in .text section in /home/shorne/work/openrisc/3eb9f9d0f6d8274b2d19753c006bd83f7d536e3c/output/host/or1k-buildroot-linux-uclibc/sysroot/lib/libc.so.This is caused by GOT code having a limit of 64k. In OpenRISC this looks to be the only relocation code pattern to be limited to 64k.
This patch allows specifying a new option -mcmodel=large which can be used to generate 2 more instructions to construct 32-bit addresses for up to 4G GOTs.
gcc/ChangeLog:
PR 99783
*config/or1k/or1k-opts.h: New file.
*config/or1k/or1k.c(or1k_legitimize_address_1,print_reloc):
Support generating gotha relocations if -mcmodel=large is specified.
*config/or1k/or1k.h(TARGET_CMODEL_SMALL,TARGET_CMODEL_LARGE):
New macros.
* config/or1k/or1k.opt (mcmodel=): New option.
* doc/invoke.text (OpenRISC Options): Document mcmodel.
Quoted 8 lines
Quoted 2 lines
Note you're the port maintainer, so you don't need approval to commit this to GCC.
Thanks for confirming, I know that after Segher told me with some previous patches. :)
For this series Giulio picked it up and tested it. We found some issues and I am going to fix with a V2. I should commit that version, after the binutils changes are in.
Sounds good. Thanks. Given we're making gcc-11 RCs you'll have to decide if its worth the risk of introducing those changes this late in the cycle is safe or not.
jeff