Relocations installed by the BPF ELF backend were sometimes incorrectly
adding the symbol value to the relocation entry addend, when the correct
relocation value was already stored in the addend. This could lead to a
relocation effectively adding the symbol value twice.
Fix that by making bpf_elf_generic_reloc () more similar to the flow of
bfd_install_relocation in the case where howto->install_addend is set,
which is how it ought to behave.
bfd/
* bpf-reloc.def (R_BPF_64_ABS32, R_BPF_64_ABS64)
(R_BPF_64_NODYLD32): Set partial_inplace to true.
* elf64-bpf.c (bpf_elf_generic_reloc): Do not include the value
of the symbol when installing relocation. Copy some additional
logic from bfd_elf_generic_reloc.
gas/
* testsuite/gas/bpf/bpf.exp: Run new test.
* testsuite/gas/bpf/elf-relo-1.d: New.
* testsuite/gas/bpf/elf-relo-1.s: New.
This patch adds linker support to patch R_BPF_64_NODYLD32 relocations.
The implementation was based on comments and code in LLVM, as the GNU
toolchain does not uses this relocation type.
Adds two new external authors to etc/update-copyright.py to cover
bfd/ax_tls.m4, and adds gprofng to dirs handled automatically, then
updates copyright messages as follows:
1) Update cgen/utils.scm emitted copyrights.
2) Run "etc/update-copyright.py --this-year" with an extra external
author I haven't committed, 'Kalray SA.', to cover gas testsuite
files (which should have their copyright message removed).
3) Build with --enable-maintainer-mode --enable-cgen-maint=yes.
4) Check out */po/*.pot which we don't update frequently.
This patch adds support for the V4 BPF instruction jal/gotol, which is
like ja/goto but it supports a signed 32-bit PC-relative (in number of
64-bit words minus one) target operand instead of the 16-bit signed
operand of the other instruction. This greatly increases the jump
range in BPF programs.
Tested in bpf-unkown-none.
bfd/ChangeLog:
2023-07-24 Jose E. Marchesi <jose.marchesi@oracle.com>
* reloc.c: New reloc BFD_RELOC_BPF_DISPCALL32.
* elf64-bpf.c (bpf_reloc_type_lookup): Handle the new reloc.
* libbfd.h (bfd_reloc_code_real_names): Regenerate.
gas/ChangeLog:
2023-07-24 Jose E. Marchesi <jose.marchesi@oracle.com>
* config/tc-bpf.c (struct bpf_insn): New field `id'.
(md_assemble): Save the ids of successfully parsed instructions
and use the new BFD_RELOC_BPF_DISPCALL32 whenever appropriate.
(md_apply_fix): Adapt to the new BFD reloc.
* testsuite/gas/bpf/jump.s: Test JAL.
* testsuite/gas/bpf/jump.d: Likewise.
* testsuite/gas/bpf/jump-pseudoc.d: Likewise.
* testsuite/gas/bpf/jump-be.d: Likewise.
* testsuite/gas/bpf/jump-be-pseudoc.d: Likewise.
* doc/c-bpf.texi (BPF Instructions): Document new instruction
jal/gotol.
Document new operand type disp32.
include/ChangeLog:
2023-07-24 Jose E. Marchesi <jose.marchesi@oracle.com>
* opcode/bpf.h (enum bpf_insn_id): Add entry BPF_INSN_JAL.
(enum bpf_insn_id): Remove spurious entry BPF_INSN_CALLI.
opcodes/ChangeLog:
2023-07-23 Jose E. Marchesi <jose.marchesi@oracle.com>
* bpf-opc.c (bpf_opcodes): Add entry for jal.
CGEN is cool, but the BPF architecture is simply too bizarre for it.
The weird way of BPF to handle endianness in instruction encoding, the
weird C-like alternative assembly syntax, the weird abuse of
multi-byte (or infra-byte) instruction fields as opcodes, the unusual
presence of opcodes beyond the first 32-bits of some instructions, are
all examples of what makes it a PITA to continue using CGEN for this
port. The bpf.cpu file is becoming so complex and so nested with
p-macros that it is very difficult to read, and quite challenging to
update. Also, every time we are forced to change something in CGEN to
accommodate BPF requirements (which is often) we have to do extensive
testing to make sure we do not break any other target using CGEN.
This is getting un-maintenable.
So I have decided to bite the bullet and revamp/rewrite the port so it
no longer uses CGEN. Overall, this involved:
* To remove the cpu/bpf.{cpu,opc} descriptions.
* To remove the CGEN generated files.
* To replace the CGEN generated opcodes table with a new hand-written
opcodes table for BPF.
* To replace the CGEN generated disassembler wih a new disassembler
that uses the new opcodes.
* To replace the CGEN generated assembler with a new assembler that uses the
new opcodes.
* To replace the CGEN generated simulator with a new simulator that uses the
new opcodes. [This is pushed in GDB in another patch.]
* To adapt the build systems to the new situation.
Additionally, this patch introduces some extensions and improvements:
* A new BPF relocation BPF_RELOC_BPF_DISP16 plus corresponding ELF
relocation R_BPF_GNU_64_16 are added to the BPF BFD port. These
relocations are used for section-relative 16-bit offsets used in
load/store instructions.
* The disassembler now has support for the "pseudo-c" assembly syntax of
BPF. What dialect to use when disassembling is controlled by a command
line option.
* The disassembler now has support for dumping instruction immediates in
either octal, hexadecimal or decimal. The used output base is controlled
by a new command-line option.
* The GAS BPF test suite has been re-structured and expanded in order to
test the disassembler pseudoc syntax support. Minor bugs have been also
fixed there. The assembler generic tests that were disabled for bpf-*-*
targets due to the previous implementation of pseudoc syntax are now
re-enabled. Additional tests have been added to test the new features of
the assembler. .dump files are no longer used.
* The linker BPF test suite has been adapted to the command line options
used by the new disassembler.
The result is very satisfactory. This patchs adds 3448 lines of code
and removes 10542 lines of code.
Tested in:
* Target bpf-unknown-none with 64-bit little-endian host and 32-bit
little-endian host.
* Target x86-64-linux-gnu with --enable-targets=all
Note that I have not tested in a big-endian host yet. I will do so
once this lands upstream so I can use the GCC compiler farm.
I have not included ChangeLog entries in this patch: these would be
massive and not very useful, considering this is pretty much a rewrite
of the port. I beg the indulgence of the global maintainers.
The newer update-copyright.py fixes file encoding too, removing cr/lf
on binutils/bfdtest2.c and ld/testsuite/ld-cygwin/exe-export.exp, and
embedded cr in binutils/testsuite/binutils-all/ar.exp string match.
The result of running etc/update-copyright.py --this-year, fixing all
the files whose mode is changed by the script, plus a build with
--enable-maintainer-mode --enable-cgen-maint=yes, then checking
out */po/*.pot which we don't update frequently.
The copy of cgen was with commit d1dd5fcc38ead reverted as that commit
breaks building of bfp opcodes files.
The bpf reloc howtos are a bit weird, using bitpos to specify an
offset from r_offset that is outside the size of the reloc as given by
howto.size. That means bfd_get_reloc_size gives the wrong answer for
range checking, and thus bfd_reloc_offset_in_range can't be used.
* elf64-bpf.c (bpf_elf_generic_reloc): Handle bitpos offset reloc
range checking.
The testcase in this PR tickled two bugs fixed here. output_bfd is
NULL when a reloc special_function is called for final linking and
when called from bfd_generic_get_relocated_section_contents. Clearly
using output_bfd is wrong as it results in segfaults. Not only that,
the endianness of the reloc field really should be that of the input.
The second bug was not checking that the entire reloc field was
contained in the section contents.
PR 28055
* elf64-bpf.c (bpf_elf_generic_reloc): Use correct bfd for bfd_put
and bfd_put_32 calls. Correct section limit checks.
When performing DISP{16,32} relocations, the eBPF ELF backend linker
needs to convert the relocation from an address into a signed number
of 64-bit words (minus one) to jump.
Because of this unsigned-to-signed conversion, special care needs to
be taken when dividing to ensure the sign bits remain correct.
Otherwise, a false relocation overflow error can be triggered.
bfd/ChangeLog
2020-08-07 David Faust <david.faust@oracle.com>
* elf64-bpf.c (bpf_elf_relocate_section): Ensure signed division for
DISP16 and DISP32 relocations.
ld/ChangeLog
2020-08-07 David Faust <david.faust@oracle.com>
* testsuite/ld-bpf/call-3.s: New file.
* testsuite/ld-bpf/call-3.d: Likewise.
The eBPF ELF backend was not properly recording relocation addends
during installation, nor reading and applying them when performing
the final relocation. This lead to various issues with incorrect
relocations.
These issues are fixed with a new howto special function to install
the relocations, and updates to bpf_elf_relocate_section to read and
use the addends as recorded in the input_bfd.
bfd/ChangeLog
2020-08-05 David Faust <david.faust@oracle.com>
* elf64-bpf.c (bpf_elf_generic_reloc): New function.
(bpf_elf_howto_table): Use it here.
(bpf_elf_relocate_section): Use addends recorded in input_bfd for
instruction and data relocations.
ld/ChangeLog
2020-08-05 David Faust <david.faust@oracle.com>
* testsuite/ld-bpf/call-2.s: New file.
* testsuite/ld-bpf/call-2.d: Likewise.
* testsuite/ld-bpf/reloc-data-be.d: Likewise.
* testsuite/ld-bpf/reloc-data-le.d: Likewise.
* testsuite/ld-bpf/reloc-data.s: Likewise.
* testsuite/ld-bpf/reloc-insn-external-be.d: Likewise.
* testsuite/ld-bpf/reloc-insn-external-le.d: Likewise.
* testsuite/ld-bpf/reloc-insn-external.s: Likewise.
* testsuite/ld-bpf/reloc-insn32-be.d: Likewise.
* testsuite/ld-bpf/reloc-insn32-le.d: Likewise.
* testsuite/ld-bpf/reloc-insn32.s: Likewise.
* testsuite/ld-bpf/reloc-insn64-be.d: Likewise.
* testsuite/ld-bpf/reloc-insn64-le.d: Likewise.
* testsuite/ld-bpf/reloc-insn64.s: Likewise.
This patch changes the eBPF linker to provide a relocate_section
function instead of relying on using special functions in relocation
howtos.
Tested in x86_64 host.
No regressions.
bfd/ChangeLog:
2019-08-07 Jose E. Marchesi <jose.marchesi@oracle.com>
* elf64-bpf.c (bpf_elf_relocate_section): New function.
(bpf_elf_insn_disp_reloc): Delete function.
(elf_backend_relocate_section): Define.
