summaryrefslogtreecommitdiff
path: root/scripts/mod/modpost.c
AgeCommit message (Collapse)Author
2024-01-31modpost: avoid using the alias attributeMasahiro Yamada
Aiden Leong reported modpost fails to build on macOS since commit 16a473f60edc ("modpost: inform compilers that fatal() never returns"): scripts/mod/modpost.c:93:21: error: aliases are not supported on darwin Nathan's research indicates that Darwin seems to support weak aliases at least [1]. Although the situation might be improved in future Clang versions, we can achieve a similar outcome without relying on it. This commit makes fatal() a macro of error() + exit(1) in modpost.h, as compilers recognize that exit() never returns. [1]: https://github.com/llvm/llvm-project/issues/71001 Fixes: 16a473f60edc ("modpost: inform compilers that fatal() never returns") Reported-by: Aiden Leong <aiden.leong@aibsd.com> Closes: https://lore.kernel.org/all/d9ac2960-6644-4a87-b5e4-4bfb6e0364a8@aibsd.com/ Signed-off-by: Masahiro Yamada <masahiroy@kernel.org>
2024-01-28modpost: Add '.ltext' and '.ltext.*' to TEXT_SECTIONSNathan Chancellor
After the linked LLVM change, building ARCH=um defconfig results in a segmentation fault in modpost. Prior to commit a23e7584ecf3 ("modpost: unify 'sym' and 'to' in default_mismatch_handler()"), there was a warning: WARNING: modpost: vmlinux.o(__ex_table+0x88): Section mismatch in reference to the .ltext:(unknown) WARNING: modpost: The relocation at __ex_table+0x88 references section ".ltext" which is not in the list of authorized sections. If you're adding a new section and/or if this reference is valid, add ".ltext" to the list of authorized sections to jump to on fault. This can be achieved by adding ".ltext" to OTHER_TEXT_SECTIONS in scripts/mod/modpost.c. The linked LLVM change moves global objects to the '.ltext' (and '.ltext.*' with '-ffunction-sections') sections with '-mcmodel=large', which ARCH=um uses. These sections should be handled just as '.text' and '.text.*' are, so add them to TEXT_SECTIONS. Cc: stable@vger.kernel.org Closes: https://github.com/ClangBuiltLinux/linux/issues/1981 Link: https://github.com/llvm/llvm-project/commit/4bf8a688956a759b7b6b8d94f42d25c13c7af130 Signed-off-by: Nathan Chancellor <nathan@kernel.org> Signed-off-by: Masahiro Yamada <masahiroy@kernel.org>
2024-01-19Merge tag 'loongarch-6.8' of ↵Linus Torvalds
git://git.kernel.org/pub/scm/linux/kernel/git/chenhuacai/linux-loongson Pull LoongArch updates from Huacai Chen: - Raise minimum clang version to 18.0.0 - Enable initial Rust support for LoongArch - Add built-in dtb support for LoongArch - Use generic interface to support crashkernel=X,[high,low] - Some bug fixes and other small changes - Update the default config file. * tag 'loongarch-6.8' of git://git.kernel.org/pub/scm/linux/kernel/git/chenhuacai/linux-loongson: (22 commits) MAINTAINERS: Add BPF JIT for LOONGARCH entry LoongArch: Update Loongson-3 default config file LoongArch: BPF: Prevent out-of-bounds memory access LoongArch: BPF: Support 64-bit pointers to kfuncs LoongArch: Fix definition of ftrace_regs_set_instruction_pointer() LoongArch: Use generic interface to support crashkernel=X,[high,low] LoongArch: Fix and simplify fcsr initialization on execve() LoongArch: Let cores_io_master cover the largest NR_CPUS LoongArch: Change SHMLBA from SZ_64K to PAGE_SIZE LoongArch: Add a missing call to efi_esrt_init() LoongArch: Parsing CPU-related information from DTS LoongArch: dts: DeviceTree for Loongson-2K2000 LoongArch: dts: DeviceTree for Loongson-2K1000 LoongArch: dts: DeviceTree for Loongson-2K0500 LoongArch: Allow device trees be built into the kernel dt-bindings: interrupt-controller: loongson,liointc: Fix dtbs_check warning for interrupt-names dt-bindings: interrupt-controller: loongson,liointc: Fix dtbs_check warning for reg-names dt-bindings: loongarch: Add Loongson SoC boards compatibles dt-bindings: loongarch: Add CPU bindings for LoongArch LoongArch: Enable initial Rust support ...
2024-01-17modpost: Ignore relaxation and alignment marker relocs on LoongArchWANG Xuerui
With recent trunk versions of binutils and gcc, alignment directives are represented with R_LARCH_ALIGN relocs on LoongArch, which is necessary for the linker to maintain alignment requirements during its relaxation passes. And even though the kernel is built with relaxation disabled, so far a small number of R_LARCH_RELAX marker relocs are still emitted as part of la.* pseudo instructions in assembly. These two kinds of relocs do not refer to symbols, which can trip up modpost's section mismatch checks, because the r_offset of said relocs can be zero or any other meaningless value, eventually leading to a `from == NULL` condition in default_mismatch_handler and SIGSEGV. As the two kinds of relocs are not concerned with symbols, just ignore them for section mismatch check purposes. Signed-off-by: WANG Xuerui <git@xen0n.name> Signed-off-by: Huacai Chen <chenhuacai@loongson.cn>
2023-12-10modpost: remove unreachable code after fatal()Masahiro Yamada
Now compilers can recognize fatal() never returns. While GCC 4.5 dropped support for -Wunreachable-code, Clang is capable of detecting the unreachable code. $ make HOSTCC=clang HOSTCFLAGS=-Wunreachable-code-return [snip] HOSTCC scripts/mod/modpost.o scripts/mod/modpost.c:520:11: warning: 'return' will never be executed [-Wunreachable-code-return] return 0; ^ scripts/mod/modpost.c:477:10: warning: 'return' will never be executed [-Wunreachable-code-return] return 0; ^ 2 warnings generated. Signed-off-by: Masahiro Yamada <masahiroy@kernel.org> Reviewed-by: Nathan Chancellor <nathan@kernel.org>
2023-12-10modpost: remove unneeded initializer in section_rel()Masahiro Yamada
This initializer was added to avoid -Wmaybe-uninitialized (gcc) and -Wsometimes-uninitialized (clang) warnings. Now that compilers recognize fatal() never returns, it is unneeded. Signed-off-by: Masahiro Yamada <masahiroy@kernel.org> Reviewed-by: Nathan Chancellor <nathan@kernel.org>
2023-12-10modpost: inform compilers that fatal() never returnsMasahiro Yamada
The function fatal() never returns because modpost_log() calls exit(1) when LOG_FATAL is passed. Inform compilers of this fact so that unreachable code flow can be identified at compile time. Signed-off-by: Masahiro Yamada <masahiroy@kernel.org> Reviewed-by: Nathan Chancellor <nathan@kernel.org>
2023-12-10modpost: move __attribute__((format(printf, 2, 3))) to modpost.hMasahiro Yamada
This attribute must be added to the function declaration in a header for comprehensive checking of all the callsites. Fixes: 6d9a89ea4b06 ("kbuild: declare the modpost error functions as printf like") Signed-off-by: Masahiro Yamada <masahiroy@kernel.org> Reviewed-by: Nathan Chancellor <nathan@kernel.org>
2023-11-16modpost: fix section mismatch message for RELAMasahiro Yamada
The section mismatch check prints a bogus symbol name on some architectures. [test code] #include <linux/init.h> int __initdata foo; int get_foo(void) { return foo; } If you compile it with GCC for riscv or loongarch, modpost will show an incorrect symbol name: WARNING: modpost: vmlinux: section mismatch in reference: get_foo+0x8 (section: .text) -> done (section: .init.data) To get the correct symbol address, the st_value must be added. This issue has never been noticed since commit 93684d3b8062 ("kbuild: include symbol names in section mismatch warnings") presumably because st_value becomes zero on most architectures when the referenced symbol is looked up. It is not true for riscv or loongarch, at least. With this fix, modpost will show the correct symbol name: WARNING: modpost: vmlinux: section mismatch in reference: get_foo+0x8 (section: .text) -> foo (section: .init.data) Signed-off-by: Masahiro Yamada <masahiroy@kernel.org> Reviewed-by: Nick Desaulniers <ndesaulniers@google.com>
2023-10-28modpost: squash ALL_{INIT,EXIT}_TEXT_SECTIONS to ALL_TEXT_SECTIONSMasahiro Yamada
ALL_INIT_TEXT_SECTIONS and ALL_EXIT_TEXT_SECTIONS are only used in the macro definition of ALL_TEXT_SECTIONS. Signed-off-by: Masahiro Yamada <masahiroy@kernel.org>
2023-10-28modpost: merge sectioncheck table entries regarding init/exit sectionsMasahiro Yamada
Check symbol references from normal sections to init/exit sections in a single entry. Signed-off-by: Masahiro Yamada <masahiroy@kernel.org>
2023-10-28modpost: use ALL_INIT_SECTIONS for the section check from DATA_SECTIONSMasahiro Yamada
ALL_INIT_SECTIONS is defined as follows: #define ALL_INIT_SECTIONS INIT_SECTIONS, ALL_XXXINIT_SECTIONS Signed-off-by: Masahiro Yamada <masahiroy@kernel.org>
2023-10-28modpost: disallow the combination of EXPORT_SYMBOL and __meminit*Masahiro Yamada
Theoretically, we could export conditionally-discarded code sections, such as .meminit*, if all the users can become modular under a certain condition. However, that would be difficult to control and such a tricky case has never occurred. Signed-off-by: Masahiro Yamada <masahiroy@kernel.org>
2023-10-28modpost: remove EXIT_SECTIONS macroMasahiro Yamada
ALL_EXIT_SECTIONS and EXIT_SECTIONS are the same. Remove the latter. Signed-off-by: Masahiro Yamada <masahiroy@kernel.org>
2023-10-28modpost: remove MEM_INIT_SECTIONS macroMasahiro Yamada
ALL_XXXINIT_SECTIONS and MEM_INIT_SECTIONS are the same. Remove the latter. Signed-off-by: Masahiro Yamada <masahiroy@kernel.org>
2023-10-28modpost: remove more symbol patterns from the section check whitelistMasahiro Yamada
These symbol patterns were whitelisted to allow them to reference to functions with the old __devinit and __devexit annotations. We stopped doing this a long time ago, for example, commit 6f039790510f ("Drivers: scsi: remove __dev* attributes.") remove those annotations from the scsi drivers. Keep *_ops, *_probe, and *_console, otherwise they will really cause section mismatch warnings. Signed-off-by: Masahiro Yamada <masahiroy@kernel.org>
2023-10-28modpost: disallow *driver to reference .meminit* sectionsMasahiro Yamada
Drivers must not reference .meminit* sections, which are discarded when CONFIG_MEMORY_HOTPLUG=n. The reason for whitelisting "*driver" in the section mismatch check was to allow drivers to reference symbols annotated as __devinit or __devexit that existed in the past. Those annotations were removed by the following commits: - 54b956b90360 ("Remove __dev* markings from init.h") - 92e9e6d1f984 ("modpost.c: Stop checking __dev* section mismatches") Remove the stale whitelist. Signed-off-by: Masahiro Yamada <masahiroy@kernel.org>
2023-10-28linux/init: remove __memexit* annotationsMasahiro Yamada
We have never used __memexit, __memexitdata, or __memexitconst. These were unneeded. Signed-off-by: Masahiro Yamada <masahiroy@kernel.org> Acked-by: Arnd Bergmann <arnd@arndb.de>
2023-10-28modpost: remove ALL_EXIT_DATA_SECTIONS macroMasahiro Yamada
This is unused. Signed-off-by: Masahiro Yamada <masahiroy@kernel.org>
2023-10-18modpost: factor out the common boilerplate of section_rel(a)Masahiro Yamada
The first few lines of section_rel() and section_rela() are the same. They both retrieve the index of the section to which the relocaton applies, and skip known-good sections. This common code should be moved to check_sec_ref(). Avoid ugly casts when computing 'start' and 'stop', and also make the Elf_Rel and Elf_Rela pointers const. Signed-off-by: Masahiro Yamada <masahiroy@kernel.org> Reviewed-by: Nick Desaulniers <ndesaulniers@google.com>
2023-10-18modpost: refactor check_sec_ref()Masahiro Yamada
We can replace &elf->sechdrs[i] with &sechdrs[i] to slightly shorten the code because we already have the local variable 'sechdrs'. However, defining 'sechdr' instead shortens the code further. Signed-off-by: Masahiro Yamada <masahiroy@kernel.org> Reviewed-by: Nick Desaulniers <ndesaulniers@google.com>
2023-10-18modpost: define TO_NATIVE() using bswap_* functionsMasahiro Yamada
The current TO_NATIVE() has some limitations: 1) You cannot cast the argument. 2) You cannot pass a variable marked as 'const'. 3) Passing an array is a bug, but it is not detected. Impelement TO_NATIVE() using bswap_*() functions. These are GNU extensions. If we face portability issues, we can port the code from include/uapi/linux/swab.h. With this change, get_rel_type_and_sym() can be simplified by casting the arguments directly. Signed-off-by: Masahiro Yamada <masahiroy@kernel.org> Reviewed-by: Nick Desaulniers <ndesaulniers@google.com>
2023-10-03modpost: Optimize symbol search from linear to binary searchJack Brennen
Modify modpost to use binary search for converting addresses back into symbol references. Previously it used linear search. This change saves a few seconds of wall time for defconfig builds, but can save several minutes on allyesconfigs. Before: $ make LLVM=1 -j128 allyesconfig vmlinux -s KCFLAGS="-Wno-error" $ time scripts/mod/modpost -M -m -a -N -o vmlinux.symvers vmlinux.o 198.38user 1.27system 3:19.71elapsed After: $ make LLVM=1 -j128 allyesconfig vmlinux -s KCFLAGS="-Wno-error" $ time scripts/mod/modpost -M -m -a -N -o vmlinux.symvers vmlinux.o 11.91user 0.85system 0:12.78elapsed Signed-off-by: Jack Brennen <jbrennen@google.com> Tested-by: Nick Desaulniers <ndesaulniers@google.com> Signed-off-by: Masahiro Yamada <masahiroy@kernel.org>
2023-10-01modpost: Don't let "driver"s reference .exit.*Uwe Kleine-König
Drivers must not reference functions marked with __exit as these likely are not available when the code is built-in. There are few creative offenders uncovered for example in ARCH=amd64 allmodconfig builds. So only trigger the section mismatch warning for W=1 builds. The dual rule that drivers must not reference .init.* is implemented since commit 0db252452378 ("modpost: don't allow *driver to reference .init.*") which however missed that .exit.* should be handled in the same way. Thanks to Masahiro Yamada and Arnd Bergmann who gave valuable hints to find this improvement. Signed-off-by: Uwe Kleine-König <u.kleine-koenig@pengutronix.de> Signed-off-by: Masahiro Yamada <masahiroy@kernel.org>
2023-09-13Merge tag 'parisc-for-6.6-rc2' of ↵Linus Torvalds
git://git.kernel.org/pub/scm/linux/kernel/git/deller/parisc-linux Pull parisc architecture fixes from Helge Deller: - fix reference to exported symbols for parisc64 [Masahiro Yamada] - Block-TLB (BTLB) support on 32-bit CPUs - sparse and build-warning fixes * tag 'parisc-for-6.6-rc2' of git://git.kernel.org/pub/scm/linux/kernel/git/deller/parisc-linux: linux/export: fix reference to exported functions for parisc64 parisc: BTLB: Initialize BTLB tables at CPU startup parisc: firmware: Simplify calling non-PA20 functions parisc: BTLB: _edata symbol has to be page aligned for BTLB support parisc: BTLB: Add BTLB insert and purge firmware function wrappers parisc: BTLB: Clear possibly existing BTLB entries parisc: Prepare for Block-TLB support on 32-bit kernel parisc: shmparam.h: Document aliasing requirements of PA-RISC parisc: irq: Make irq_stack_union static to avoid sparse warning parisc: drivers: Fix sparse warning parisc: iosapic.c: Fix sparse warnings parisc: ccio-dma: Fix sparse warnings parisc: sba-iommu: Fix sparse warnigs parisc: sba: Fix compile warning wrt list of SBA devices parisc: sba_iommu: Fix build warning if procfs if disabled
2023-09-12linux/export: fix reference to exported functions for parisc64Masahiro Yamada
John David Anglin reported parisc has been broken since commit ddb5cdbafaaa ("kbuild: generate KSYMTAB entries by modpost"). Like ia64, parisc64 uses a function descriptor. The function references must be prefixed with P%. Also, symbols prefixed $$ from the library have the symbol type STT_LOPROC instead of STT_FUNC. They should be handled as functions too. Fixes: ddb5cdbafaaa ("kbuild: generate KSYMTAB entries by modpost") Reported-by: John David Anglin <dave.anglin@bell.net> Tested-by: John David Anglin <dave.anglin@bell.net> Tested-by: Helge Deller <deller@gmx.de> Closes: https://lore.kernel.org/linux-parisc/1901598a-e11d-f7dd-a5d9-9a69d06e6b6e@bell.net/T/#u Signed-off-by: Masahiro Yamada <masahiroy@kernel.org> Signed-off-by: Helge Deller <deller@gmx.de>
2023-09-01modpost: Skip .llvm.call-graph-profile section checkDenis Nikitin
.llvm.call-graph-profile section is added by clang when the kernel is built with profiles (e.g. -fprofile-sample-use= or -fprofile-use=). Note that .llvm.call-graph-profile intentionally uses REL relocations to decrease the object size, for more details see https://reviews.llvm.org/D104080. The section contains edge information derived from text sections, so .llvm.call-graph-profile itself doesn't need more analysis as the text sections have been analyzed. This change fixes the kernel build with clang and a sample profile which currently fails with: "FATAL: modpost: Please add code to calculate addend for this architecture" Signed-off-by: Denis Nikitin <denik@chromium.org> Reviewed-by: Nick Desaulniers <ndesaulniers@google.com> Reviewed-by: Fangrui Song <maskray@google.com> Signed-off-by: Masahiro Yamada <masahiroy@kernel.org>
2023-07-31modpost: remove ElF_Rela variables from for-loop in section_rel(a)Masahiro Yamada
Remove the Elf_Rela variables used in the for-loop in section_rel(). This makes the code consistent; section_rel() only uses Elf_Rel, section_rela() only uses Elf_Rela. Signed-off-by: Masahiro Yamada <masahiroy@kernel.org>
2023-07-31modpost: clean up MIPS64 little endian relocation codeMasahiro Yamada
MIPS64 little endian target has an odd encoding of r_info. This commit makes the special handling less ugly. It is still ugly, but #if conditionals will go away, at least. Signed-off-by: Masahiro Yamada <masahiroy@kernel.org>
2023-07-31modpost: pass r_type to addend_*_rel()Masahiro Yamada
All of addend_*_rel() need the Elf_Rela pointer just for calculating ELF_R_TYPE(r->r_info). You can do it on the caller to de-duplicate the code. Signed-off-by: Masahiro Yamada <masahiroy@kernel.org>
2023-07-31modpost: change return type of addend_*_rel()Masahiro Yamada
Now that none of addend_*_rel() returns a meaningful value (the return value is always 0), change all of them to return the value of r_addend. Signed-off-by: Masahiro Yamada <masahiroy@kernel.org>
2023-07-25linux/export.h: make <linux/export.h> independent of CONFIG_MODULESMasahiro Yamada
Currently, all files with EXPORT_SYMBOL() are rebuilt when CONFIG_MODULES is flipped due to <linux/export.h> depending on CONFIG_MODULES. Now that modpost can make a final decision about export symbols, <linux/export.h> does not need to make EXPORT_SYMBOL() no-op. Instead, modpost can skip emitting KSYMTAB when CONFIG_MODULES is unset. This commit will reduce the number of recompilation when CONFIG_MODULES is toggled. Signed-off-by: Masahiro Yamada <masahiroy@kernel.org>
2023-07-01Merge tag 'kbuild-v6.5' of ↵Linus Torvalds
git://git.kernel.org/pub/scm/linux/kernel/git/masahiroy/linux-kbuild Pull Kbuild updates from Masahiro Yamada: - Remove the deprecated rule to build *.dtbo from *.dts - Refactor section mismatch detection in modpost - Fix bogus ARM section mismatch detections - Fix error of 'make gtags' with O= option - Add Clang's target triple to KBUILD_CPPFLAGS to fix a build error with the latest LLVM version - Rebuild the built-in initrd when KBUILD_BUILD_TIMESTAMP is changed - Ignore more compiler-generated symbols for kallsyms - Fix 'make local*config' to handle the ${CONFIG_FOO} form in Makefiles - Enable more kernel-doc warnings with W=2 - Refactor <linux/export.h> by generating KSYMTAB data by modpost - Deprecate <asm/export.h> and <asm-generic/export.h> - Remove the EXPORT_DATA_SYMBOL macro - Move the check for static EXPORT_SYMBOL back to modpost, which makes the build faster - Re-implement CONFIG_TRIM_UNUSED_KSYMS with one-pass algorithm - Warn missing MODULE_DESCRIPTION when building modules with W=1 - Make 'make clean' robust against too long argument error - Exclude more objects from GCOV to fix CFI failures with GCOV - Allow 'make modules_install' to install modules.builtin and modules.builtin.modinfo even when CONFIG_MODULES is disabled - Include modules.builtin and modules.builtin.modinfo in the linux-image Debian package even when CONFIG_MODULES is disabled - Revive "Entering directory" logging for the latest Make version * tag 'kbuild-v6.5' of git://git.kernel.org/pub/scm/linux/kernel/git/masahiroy/linux-kbuild: (72 commits) modpost: define more R_ARM_* for old distributions kbuild: revive "Entering directory" for Make >= 4.4.1 kbuild: set correct abs_srctree and abs_objtree for package builds scripts/mksysmap: Ignore prefixed KCFI symbols kbuild: deb-pkg: remove the CONFIG_MODULES check in buildeb kbuild: builddeb: always make modules_install, to install modules.builtin* modpost: continue even with unknown relocation type modpost: factor out Elf_Sym pointer calculation to section_rel() modpost: factor out inst location calculation to section_rel() kbuild: Disable GCOV for *.mod.o kbuild: Fix CFI failures with GCOV kbuild: make clean rule robust against too long argument error script: modpost: emit a warning when the description is missing kbuild: make modules_install copy modules.builtin(.modinfo) linux/export.h: rename 'sec' argument to 'license' modpost: show offset from symbol for section mismatch warnings modpost: merge two similar section mismatch warnings kbuild: implement CONFIG_TRIM_UNUSED_KSYMS without recursion modpost: use null string instead of NULL pointer for default namespace modpost: squash sym_update_namespace() into sym_add_exported() ...
2023-06-29modpost: define more R_ARM_* for old distributionsMasahiro Yamada
On CentOS 7, the following build error occurs. scripts/mod/modpost.c: In function 'addend_arm_rel': scripts/mod/modpost.c:1312:7: error: 'R_ARM_MOVW_ABS_NC' undeclared (first use in this function); did you mean 'R_ARM_THM_ABS5'? case R_ARM_MOVW_ABS_NC: ^~~~~~~~~~~~~~~~~ R_ARM_THM_ABS5 scripts/mod/modpost.c:1312:7: note: each undeclared identifier is reported only once for each function it appears in scripts/mod/modpost.c:1313:7: error: 'R_ARM_MOVT_ABS' undeclared (first use in this function); did you mean 'R_ARM_THM_ABS5'? case R_ARM_MOVT_ABS: ^~~~~~~~~~~~~~ R_ARM_THM_ABS5 scripts/mod/modpost.c:1326:7: error: 'R_ARM_THM_MOVW_ABS_NC' undeclared (first use in this function); did you mean 'R_ARM_THM_ABS5'? case R_ARM_THM_MOVW_ABS_NC: ^~~~~~~~~~~~~~~~~~~~~ R_ARM_THM_ABS5 scripts/mod/modpost.c:1327:7: error: 'R_ARM_THM_MOVT_ABS' undeclared (first use in this function); did you mean 'R_ARM_THM_ABS5'? case R_ARM_THM_MOVT_ABS: ^~~~~~~~~~~~~~~~~~ R_ARM_THM_ABS5 Fixes: 12ca2c67d742 ("modpost: detect section mismatch for R_ARM_{MOVW_ABS_NC,MOVT_ABS}") Fixes: cd1824fb7a37 ("modpost: detect section mismatch for R_ARM_THM_{MOVW_ABS_NC,MOVT_ABS}") Reported-by: Tetsuo Handa <penguin-kernel@I-love.SAKURA.ne.jp> Signed-off-by: Masahiro Yamada <masahiroy@kernel.org>
2023-06-25modpost: continue even with unknown relocation typeMasahiro Yamada
Currently, unknown relocation types are just skipped. The value of r_addend is only needed to get the symbol name in case is_valid_name(elf, sym) returns false. Even if we do not know how to calculate r_addend, we should continue. At worst, we will get "(unknown)" as the symbol name, but it is better than failing to detect section mismatches. Signed-off-by: Masahiro Yamada <masahiroy@kernel.org>
2023-06-25modpost: factor out Elf_Sym pointer calculation to section_rel()Masahiro Yamada
Pass the Elf_Sym pointer to addend_arm_rel() as well as to check_section_mismatch(). Signed-off-by: Masahiro Yamada <masahiroy@kernel.org> Reviewed-by: Nick Desaulniers <ndesaulniers@google.com>
2023-06-25modpost: factor out inst location calculation to section_rel()Masahiro Yamada
All the addend_*_rel() functions calculate the instruction location in the same way. Factor out the similar code to the caller. Squash reloc_location() too. Signed-off-by: Masahiro Yamada <masahiroy@kernel.org>
2023-06-24script: modpost: emit a warning when the description is missingVincenzo Palazzo
Emit a warning when the mod description is missed and only when the W=1 is enabled. Reported-by: Roland Kletzing <devzero@web.de> Link: https://bugzilla.kernel.org/show_bug.cgi?id=10770 Signed-off-by: Vincenzo Palazzo <vincenzopalazzodev@gmail.com> Tested-by: Nicolas Schier <n.schier@avm.de> Signed-off-by: Masahiro Yamada <masahiroy@kernel.org>
2023-06-22modpost: show offset from symbol for section mismatch warningsMasahiro Yamada
Currently, modpost only shows the symbol names and section names, so it repeats the same message if there are multiple relocations in the same symbol. It is common the relocation spans across multiple instructions. It is better to show the offset from the symbol. Signed-off-by: Masahiro Yamada <masahiroy@kernel.org> Reviewed-by: Nick Desaulniers <ndesaulniers@google.com>
2023-06-22modpost: merge two similar section mismatch warningsMasahiro Yamada
In case of section mismatch, modpost shows slightly different messages. For extable section mismatch: "%s(%s+0x%lx): Section mismatch in reference to the %s:%s\n" For the other cases: "%s: section mismatch in reference: %s (section: %s) -> %s (section: %s)\n" They are similar. Merge them. Signed-off-by: Masahiro Yamada <masahiroy@kernel.org> Reviewed-by: Nick Desaulniers <ndesaulniers@google.com>
2023-06-22kbuild: implement CONFIG_TRIM_UNUSED_KSYMS without recursionMasahiro Yamada
When CONFIG_TRIM_UNUSED_KSYMS is enabled, Kbuild recursively traverses the directory tree to determine which EXPORT_SYMBOL to trim. If an EXPORT_SYMBOL turns out to be unused by anyone, Kbuild begins the second traverse, where some source files are recompiled with their EXPORT_SYMBOL() tuned into a no-op. Linus stated negative opinions about this slowness in commits: - 5cf0fd591f2e ("Kbuild: disable TRIM_UNUSED_KSYMS option") - a555bdd0c58c ("Kbuild: enable TRIM_UNUSED_KSYMS again, with some guarding") We can do this better now. The final data structures of EXPORT_SYMBOL are generated by the modpost stage, so modpost can selectively emit KSYMTAB entries that are really used by modules. Commit f73edc8951b2 ("kbuild: unify two modpost invocations") is another ground-work to do this in a one-pass algorithm. With the list of modules, modpost sets sym->used if it is used by a module. modpost emits KSYMTAB only for symbols with sym->used==true. BTW, Nicolas explained why the trimming was implemented with recursion: https://lore.kernel.org/all/2o2rpn97-79nq-p7s2-nq5-8p83391473r@syhkavp.arg/ Actually, we never achieved that level of optimization where the chain reaction of trimming comes into play because: - CONFIG_LTO_CLANG cannot remove any unused symbols - CONFIG_LD_DEAD_CODE_DATA_ELIMINATION is enabled only for vmlinux, but not modules If deeper trimming is required, we need to revisit this, but I guess that is unlikely to happen. Signed-off-by: Masahiro Yamada <masahiroy@kernel.org>
2023-06-22modpost: use null string instead of NULL pointer for default namespaceMasahiro Yamada
The default namespace is the null string, "". When set, the null string "" is converted to NULL: s->namespace = namespace[0] ? NOFAIL(strdup(namespace)) : NULL; When printed, the NULL pointer is get back to the null string: sym->namespace ?: "" This saves 1 byte memory allocated for "", but loses the readability. In kernel-space, we strive to save memory, but modpost is a userspace tool used to build the kernel. On modern systems, such small piece of memory is not a big deal. Handle the namespace string as is. Signed-off-by: Masahiro Yamada <masahiroy@kernel.org> Reviewed-by: Nick Desaulniers <ndesaulniers@google.com>
2023-06-22modpost: squash sym_update_namespace() into sym_add_exported()Masahiro Yamada
Pass a set of the name, license, and namespace to sym_add_exported(). sym_update_namespace() is unneeded. Signed-off-by: Masahiro Yamada <masahiroy@kernel.org> Reviewed-by: Nick Desaulniers <ndesaulniers@google.com>
2023-06-22modpost: check static EXPORT_SYMBOL* by modpost againMasahiro Yamada
Commit 31cb50b5590f ("kbuild: check static EXPORT_SYMBOL* by script instead of modpost") moved the static EXPORT_SYMBOL* check from the mostpost to a shell script because I thought it must be checked per compilation unit to avoid false negatives. I came up with an idea to do this in modpost, against combined ELF files. The relocation entries in ELF will find the correct exported symbol even if there exist symbols with the same name in different compilation units. Again, the same sample code. Makefile: obj-y += foo1.o foo2.o foo1.c: #include <linux/export.h> static void foo(void) {} EXPORT_SYMBOL(foo); foo2.c: void foo(void) {} Then, modpost can catch it correctly. MODPOST Module.symvers ERROR: modpost: vmlinux: local symbol 'foo' was exported Signed-off-by: Masahiro Yamada <masahiroy@kernel.org> Reviewed-by: Nick Desaulniers <ndesaulniers@google.com>
2023-06-22kbuild: generate KSYMTAB entries by modpostMasahiro Yamada
Commit 7b4537199a4a ("kbuild: link symbol CRCs at final link, removing CONFIG_MODULE_REL_CRCS") made modpost output CRCs in the same way whether the EXPORT_SYMBOL() is placed in *.c or *.S. For further cleanups, this commit applies a similar approach to the entire data structure of EXPORT_SYMBOL(). The EXPORT_SYMBOL() compilation is split into two stages. When a source file is compiled, EXPORT_SYMBOL() will be converted into a dummy symbol in the .export_symbol section. For example, EXPORT_SYMBOL(foo); EXPORT_SYMBOL_NS_GPL(bar, BAR_NAMESPACE); will be encoded into the following assembly code: .section ".export_symbol","a" __export_symbol_foo: .asciz "" /* license */ .asciz "" /* name space */ .balign 8 .quad foo /* symbol reference */ .previous .section ".export_symbol","a" __export_symbol_bar: .asciz "GPL" /* license */ .asciz "BAR_NAMESPACE" /* name space */ .balign 8 .quad bar /* symbol reference */ .previous They are mere markers to tell modpost the name, license, and namespace of the symbols. They will be dropped from the final vmlinux and modules because the *(.export_symbol) will go into /DISCARD/ in the linker script. Then, modpost extracts all the information about EXPORT_SYMBOL() from the .export_symbol section, and generates the final C code: KSYMTAB_FUNC(foo, "", ""); KSYMTAB_FUNC(bar, "_gpl", "BAR_NAMESPACE"); KSYMTAB_FUNC() (or KSYMTAB_DATA() if it is data) is expanded to struct kernel_symbol that will be linked to the vmlinux or a module. With this change, EXPORT_SYMBOL() works in the same way for *.c and *.S files, providing the following benefits. [1] Deprecate EXPORT_DATA_SYMBOL() In the old days, EXPORT_SYMBOL() was only available in C files. To export a symbol in *.S, EXPORT_SYMBOL() was placed in a separate *.c file. arch/arm/kernel/armksyms.c is one example written in the classic manner. Commit 22823ab419d8 ("EXPORT_SYMBOL() for asm") removed this limitation. Since then, EXPORT_SYMBOL() can be placed close to the symbol definition in *.S files. It was a nice improvement. However, as that commit mentioned, you need to use EXPORT_DATA_SYMBOL() for data objects on some architectures. In the new approach, modpost checks symbol's type (STT_FUNC or not), and outputs KSYMTAB_FUNC() or KSYMTAB_DATA() accordingly. There are only two users of EXPORT_DATA_SYMBOL: EXPORT_DATA_SYMBOL_GPL(empty_zero_page) (arch/ia64/kernel/head.S) EXPORT_DATA_SYMBOL(ia64_ivt) (arch/ia64/kernel/ivt.S) They are transformed as follows and output into .vmlinux.export.c KSYMTAB_DATA(empty_zero_page, "_gpl", ""); KSYMTAB_DATA(ia64_ivt, "", ""); The other EXPORT_SYMBOL users in ia64 assembly are output as KSYMTAB_FUNC(). EXPORT_DATA_SYMBOL() is now deprecated. [2] merge <linux/export.h> and <asm-generic/export.h> There are two similar header implementations: include/linux/export.h for .c files include/asm-generic/export.h for .S files Ideally, the functionality should be consistent between them, but they tend to diverge. Commit 8651ec01daed ("module: add support for symbol namespaces.") did not support the namespace for *.S files. This commit shifts the essential implementation part to C, which supports EXPORT_SYMBOL_NS() for *.S files. <asm/export.h> and <asm-generic/export.h> will remain as a wrapper of <linux/export.h> for a while. They will be removed after #include <asm/export.h> directives are all replaced with #include <linux/export.h>. [3] Implement CONFIG_TRIM_UNUSED_KSYMS in one-pass algorithm (by a later commit) When CONFIG_TRIM_UNUSED_KSYMS is enabled, Kbuild recursively traverses the directory tree to determine which EXPORT_SYMBOL to trim. If an EXPORT_SYMBOL turns out to be unused by anyone, Kbuild begins the second traverse, where some source files are recompiled with their EXPORT_SYMBOL() tuned into a no-op. We can do this better now; modpost can selectively emit KSYMTAB entries that are really used by modules. Signed-off-by: Masahiro Yamada <masahiroy@kernel.org> Reviewed-by: Nick Desaulniers <ndesaulniers@google.com>
2023-06-16x86/unwind/orc: Add ELF section with ORC version identifierOmar Sandoval
Commits ffb1b4a41016 ("x86/unwind/orc: Add 'signal' field to ORC metadata") and fb799447ae29 ("x86,objtool: Split UNWIND_HINT_EMPTY in two") changed the ORC format. Although ORC is internal to the kernel, it's the only way for external tools to get reliable kernel stack traces on x86-64. In particular, the drgn debugger [1] uses ORC for stack unwinding, and these format changes broke it [2]. As the drgn maintainer, I don't care how often or how much the kernel changes the ORC format as long as I have a way to detect the change. It suffices to store a version identifier in the vmlinux and kernel module ELF files (to use when parsing ORC sections from ELF), and in kernel memory (to use when parsing ORC from a core dump+symbol table). Rather than hard-coding a version number that needs to be manually bumped, Peterz suggested hashing the definitions from orc_types.h. If there is a format change that isn't caught by this, the hashing script can be updated. This patch adds an .orc_header allocated ELF section containing the 20-byte hash to vmlinux and kernel modules, along with the corresponding __start_orc_header and __stop_orc_header symbols in vmlinux. 1: https://github.com/osandov/drgn 2: https://github.com/osandov/drgn/issues/303 Fixes: ffb1b4a41016 ("x86/unwind/orc: Add 'signal' field to ORC metadata") Fixes: fb799447ae29 ("x86,objtool: Split UNWIND_HINT_EMPTY in two") Signed-off-by: Omar Sandoval <osandov@fb.com> Signed-off-by: Peter Zijlstra (Intel) <peterz@infradead.org> Acked-by: Josh Poimboeuf <jpoimboe@kernel.org> Link: https://lkml.kernel.org/r/aef9c8dc43915b886a8c48509a12ec1b006ca1ca.1686690801.git.osandov@osandov.com
2023-06-15modpost: pass struct module pointer to check_section_mismatch()Masahiro Yamada
The next commit will use it. Signed-off-by: Masahiro Yamada <masahiroy@kernel.org> Reviewed-by: Nick Desaulniers <ndesaulniers@google.com>
2023-06-08modpost: fix off by one in is_executable_section()Dan Carpenter
The > comparison should be >= to prevent an out of bounds array access. Fixes: 52dc0595d540 ("modpost: handle relocations mismatch in __ex_table.") Signed-off-by: Dan Carpenter <dan.carpenter@linaro.org> Signed-off-by: Masahiro Yamada <masahiroy@kernel.org>
2023-06-07modpost: propagate W=1 build option to modpostMasahiro Yamada
"No build warning" is a strong requirement these days, so you must fix all issues before enabling a new warning flag. We often add a new warning to W=1 first so that the kbuild test robot blocks new breakages. This commit allows modpost to show extra warnings only when W=1 (or KBUILD_EXTRA_WARN=1) is given. Signed-off-by: Masahiro Yamada <masahiroy@kernel.org> Reviewed-by: Nick Desaulniers <ndesaulniers@google.com>
2023-06-04modpost: detect section mismatch for R_ARM_REL32Masahiro Yamada
For ARM, modpost fails to detect some types of section mismatches. [test code] .section .init.data,"aw" bar: .long 0 .section .data,"aw" .globl foo foo: .long bar - . It is apparently a bad reference, but modpost does not report anything. The test code above produces the following relocations. Relocation section '.rel.data' at offset 0xe8 contains 1 entry: Offset Info Type Sym.Value Sym. Name 00000000 00000403 R_ARM_REL32 00000000 .init.data Currently, R_ARM_REL32 is just skipped. Handle it like R_ARM_ABS32. Signed-off-by: Masahiro Yamada <masahiroy@kernel.org>