linux/tools/objtool/elf.c, branch v5.8

objtool: Provide elf_write_{insn,reloc}()

2020-06-18T15:36:33Z

This provides infrastructure to rewrite instructions; this is immediately useful for helping out with KCOV-vs-noinstr, but will also come in handy for a bunch of variable sized jump-label patches that are still on ice. Signed-off-by: Peter Zijlstra (Intel)

objtool: Clean up elf_write() condition

2020-06-18T15:36:33Z

With there being multiple ways to change the ELF data, let's more concisely track modification. Signed-off-by: Peter Zijlstra (Intel)

objtool: use gelf_getsymshndx to handle >64k sections

2020-05-15T08:35:13Z

Currently, objtool fails to load the correct section for symbols when the index is greater than SHN_LORESERVE. Use gelf_getsymshndx instead of gelf_getsym to handle >64k sections. Signed-off-by: Sami Tolvanen Signed-off-by: Peter Zijlstra (Intel) Reviewed-by: Kees Cook Link: https://lkml.kernel.org/r/20200421220843.188260-2-samitolvanen@google.com

objtool: Move the IRET hack into the arch decoder

2020-04-30T18:14:33Z

Quoting Julien: "And the other suggestion is my other email was that you don't even need to add INSN_EXCEPTION_RETURN. You can keep IRET as INSN_CONTEXT_SWITCH by default and x86 decoder lookups the symbol conaining an iret. If it's a function symbol, it can just set the type to INSN_OTHER so that it caries on to the next instruction after having handled the stack_op." Suggested-by: Julien Thierry Signed-off-by: Miroslav Benes Signed-off-by: Peter Zijlstra (Intel) Reviewed-by: Miroslav Benes Acked-by: Josh Poimboeuf Link: https://lkml.kernel.org/r/20200428191659.913283807@infradead.org

objtool: Rename elf_read() to elf_open_read()

2020-04-23T06:34:18Z

'struct elf *' handling is an open/close paradigm, make sure the naming matches that: elf_open_read() elf_write() elf_close() Acked-by: Josh Poimboeuf Signed-off-by: Ingo Molnar Cc: Borislav Petkov Cc: Linus Torvalds Cc: Peter Zijlstra Cc: Sami Tolvanen Cc: Thomas Gleixner Link: https://lore.kernel.org/r/20200422103205.61900-3-mingo@kernel.org

objtool: Constify 'struct elf *' parameters

2020-04-23T06:34:18Z

In preparation to parallelize certain parts of objtool, map out which uses of various data structures are read-only vs. read-write. As a first step constify 'struct elf' pointer passing, most of the secondary uses of it in find_symbol_*() methods are read-only. Also, while at it, better group the 'struct elf' handling methods in elf.h. Acked-by: Josh Poimboeuf Signed-off-by: Ingo Molnar Cc: Borislav Petkov Cc: Linus Torvalds Cc: Peter Zijlstra Cc: Sami Tolvanen Cc: Thomas Gleixner Link: https://lore.kernel.org/r/20200422103205.61900-2-mingo@kernel.org

objtool: Optimize !vmlinux.o again

2020-04-22T08:53:50Z

When doing kbuild tests to see if the objtool changes affected those I found that there was a measurable regression: pre post real 1m13.594 1m16.488s user 34m58.246s 35m23.947s sys 4m0.393s 4m27.312s Perf showed that for small files the increased hash-table sizes were a measurable difference. Since we already have -l "vmlinux" to distinguish between the modes, make it also use a smaller portion of the hash-tables. This flips it into a small win: real 1m14.143s user 34m49.292s sys 3m44.746s Signed-off-by: Peter Zijlstra (Intel) Reviewed-by: Miroslav Benes Reviewed-by: Alexandre Chartre Acked-by: Josh Poimboeuf Link: https://lkml.kernel.org/r/20200416115119.167588731@infradead.org Signed-off-by: Ingo Molnar

objtool: Fix off-by-one in symbol_by_offset()

2020-04-22T08:53:49Z

Sometimes, WARN_FUNC() and other users of symbol_by_offset() will associate the first instruction of a symbol with the symbol preceding it. This is because symbol->offset + symbol->len is already outside of the symbol's range. Fixes: 2a362ecc3ec9 ("objtool: Optimize find_symbol_*() and read_symbols()") Signed-off-by: Julien Thierry Reviewed-by: Miroslav Benes Signed-off-by: Josh Poimboeuf Signed-off-by: Ingo Molnar

objtool: Optimize find_rela_by_dest_range()

2020-03-25T17:28:31Z

Perf shows there is significant time in find_rela_by_dest(); this is because we have to iterate the address space per byte, looking for relocation entries. Optimize this by reducing the address space granularity. This reduces objtool on vmlinux.o runtime from 4.8 to 4.4 seconds. Signed-off-by: Peter Zijlstra (Intel) Reviewed-by: Miroslav Benes Acked-by: Josh Poimboeuf Link: https://lkml.kernel.org/r/20200324160924.861321325@infradead.org

objtool: Optimize read_sections()

2020-03-25T17:28:30Z

Perf showed that __hash_init() is a significant portion of read_sections(), so instead of doing a per section rela_hash, use an elf-wide rela_hash. Statistics show us there are about 1.1 million relas, so size it accordingly. This reduces the objtool on vmlinux.o runtime to a third, from 15 to 5 seconds. Signed-off-by: Peter Zijlstra (Intel) Reviewed-by: Miroslav Benes Acked-by: Josh Poimboeuf Link: https://lkml.kernel.org/r/20200324160924.739153726@infradead.org