+++ Matthias Maennich [06/09/19 11:32 +0100]:
As of Linux 5.3-rc7, there are 31207 [1] exported symbols in the kernel. That is a growth of roughly 1000 symbols since 4.17 (30206 [2]). There seems to be some consensus amongst kernel devs that the export surface is too large, and hard to reason about. Generally, these symbols fall in one of these categories: 1) Symbols actually meant for drivers 2) Symbols that are only exported because functionality is split over multiple modules, yet they really shouldn't be used by modules outside of their own subsystem 3) Symbols really only meant for in-tree use When module developers try to upstream their code, it regularly turns out that they are using exported symbols that they really shouldn't be using. This problem is even bigger for drivers that are currently out-of-tree, which may be using many symbols that they shouldn't be using, and that break when those symbols are removed or modified. This patch allows subsystem maintainers to partition their exported symbols into separate namespaces, and module authors to import such namespaces only when needed. This allows subsystem maintainers to more easily limit availability of these namespaced symbols to other parts of the kernel. It can also be used to partition the set of exported symbols for documentation purposes; for example, a set of symbols that is really only used for debugging could be in a "SUBSYSTEM_DEBUG" namespace. I continued the work mainly done by Martijn Coenen. Changes in v2: - Rather than adding and evaluating separate sections __knsimport_NS, use modinfo tags to declare the namespaces a module introduces. Adjust modpost and the module loader accordingly. - Also add support for reading multiple modinfo values for the same tag to allow list-like access to modinfo tags. - The macros in export.h have been cleaned up to avoid redundancy in the macro parameters (ns, nspost, nspost2). - The introduction of relative references in the ksymtab entries caused a rework of the macros to accommodate that configuration as well. - Alignment of kernel_symbol in the ksymtab needed to be fixed to allow growing the kernel_symbol struct. - Modpost does now also append the namespace suffix to the symbol entries in Module.symvers. - The configuration option MODULE_ALLOW_MISSING_NAMESPACE_IMPORTS allows relaxing the enforcement of properly declared namespace imports at module loading time. - Symbols can be collectively exported into a namespace by defining DEFAULT_SYMBOL_NAMESPACE in the corresponding Makefile. - The requirement for a very recent coccinelle spatch has been lifted by simplifying the script. - nsdeps does now ensures MODULE_IMPORT_NS statements are sorted when patching the module source files. - Some minor bugs have been addressed in nsdeps to allow it to work with modules that have more than one source file. - The RFC for the usb-storage symbols has been simplified by using DEFAULT_SYMBOL_NAMESPACE=USB_STORAGE rather than explicitly exporting each and every symbol into that new namespace. Changes in v3: - Reword the documentation for the MODULE_ALLOW_MISSING_NAMESPACE_IMPORTS option for clarification. - Fix printed required version of spatch in coccinelle script. - Adopt kbuild changes for modpost: .mod files are no longer generated in .tmp_versions. Similarely, generate the .ns_deps files in the tree along with the .mod files. Also, nsdeps now uses modules.order as source for the list modules to consider. - Add an RFC patch to introduce the namespace WATCHDOG_CORE for symbols exported in watchdog_core.c. Changes in v4: - scripts/nsdeps: - exit on first error - support out-of-tree builds O=... - scripts/export_report.pl: update for new Module.symvers format - scripts/mod/modpost: make the namespace a separate field when exporting to Module.symvers (rather than symbol.NS) - include/linux/export.h: fixed style nits - kernel/module.c: ensure namespaces are imported before taking a reference to the owner module - Documentation: document the Symbol Namespace feature and update references to Module.symvers and EXPORT_SYMBOL* Changes in v5: - Makefile: let 'nsdeps' depend on 'modules' to allow `make clean; make nsdeps` to work - scripts/nsdeps: drop 'exit on first error' again as it just makes more problems than it solves - drop the watchdog RFC patch for now This patch series was developed against v5.3-rc7.
Great work Matthias! I think this patchset is shaping up nicely. As the merge window is coming up soon, I'd like to queue this up in modules-next by the end of today to allow for some testing and "soak" time in linux-next. If there are any more complaints, please speak up. Thanks! Jessica