dd502a8107
applied to indirect function calls. Remove a data load (indirection) by modifying the text. They give the flexibility of function pointers, but with better performance. (This is especially important for cases where retpolines would otherwise be used, as retpolines can be pretty slow.) API overview: DECLARE_STATIC_CALL(name, func); DEFINE_STATIC_CALL(name, func); DEFINE_STATIC_CALL_NULL(name, typename); static_call(name)(args...); static_call_cond(name)(args...); static_call_update(name, func); x86 is supported via text patching, otherwise basic indirect calls are used, with function pointers. There's a second variant using inline code patching, inspired by jump-labels, implemented on x86 as well. The new APIs are utilized in the x86 perf code, a heavy user of function pointers, where static calls speed up the PMU handler by 4.2% (!). The generic implementation is not really excercised on other architectures, outside of the trivial test_static_call_init() self-test. Signed-off-by: Ingo Molnar <mingo@kernel.org> -----BEGIN PGP SIGNATURE----- iQJFBAABCgAvFiEEBpT5eoXrXCwVQwEKEnMQ0APhK1gFAl+EfAQRHG1pbmdvQGtl cm5lbC5vcmcACgkQEnMQ0APhK1iEAw//divHeVCJnHhV+YBbuI9ROUsERkzu8VhK O1DEmW68Fvj7pszT8NZsMjtkt97ZtxDRK7aCJiiup0eItG9qCJ8lpCLb84ZbizHV HhCbhBLrpxSvTrWlQnkgP1OkPAbtoryIjVlZzWhjye2MY8UEbVnZWyviBolbAAxH Fk1Yi56fIMu19GO+9Ohzy9E2VDnVEH1iMx5YWoLD2H88Qbq/yEMP+U2tIj8hIVKT Y/jdogihNXRIau6QB+YPfDPisdty+RHxfU7zct4Rv8cFF5ylglZB5fD34C3sUQF2 WqsaYz7zjUj9f02F8pw8hIaAT7InzArPhlNVITxf2oMfmdrNqBptnSCddZqCJLvv oDGew21k50Zcbqkv9amclpxXH5tTpRvJeqit2pz/85GMeqBRuhzHUAkCpht5YA73 qJsHWS3z+qIxKi0tDbhDJswuwa51q5sgdUUwo1uCr3wT3DGDlqNhCAZBzX14dcty 0shDSbv13TCwqAcb7asPzEoPwE15cwa+x+viGEIL901pyZKyQYjs/abDU26It3BW roWRkuVJZ9/QMdZJs1v7kaXw1L8YiKIDkBgke+xbfrDwEvvjudQkl2LUL66DB11j RJU3GyxKClvdY06SSRh/H13fqZLNKh1JZ0nPEWSTJECDFN9zcDjrDrod/7PFOcpY NAlawLoGG+s= =JvpF -----END PGP SIGNATURE----- Merge tag 'core-static_call-2020-10-12' of git://git.kernel.org/pub/scm/linux/kernel/git/tip/tip Pull static call support from Ingo Molnar: "This introduces static_call(), which is the idea of static_branch() applied to indirect function calls. Remove a data load (indirection) by modifying the text. They give the flexibility of function pointers, but with better performance. (This is especially important for cases where retpolines would otherwise be used, as retpolines can be pretty slow.) API overview: DECLARE_STATIC_CALL(name, func); DEFINE_STATIC_CALL(name, func); DEFINE_STATIC_CALL_NULL(name, typename); static_call(name)(args...); static_call_cond(name)(args...); static_call_update(name, func); x86 is supported via text patching, otherwise basic indirect calls are used, with function pointers. There's a second variant using inline code patching, inspired by jump-labels, implemented on x86 as well. The new APIs are utilized in the x86 perf code, a heavy user of function pointers, where static calls speed up the PMU handler by 4.2% (!). The generic implementation is not really excercised on other architectures, outside of the trivial test_static_call_init() self-test" * tag 'core-static_call-2020-10-12' of git://git.kernel.org/pub/scm/linux/kernel/git/tip/tip: (21 commits) static_call: Fix return type of static_call_init tracepoint: Fix out of sync data passing by static caller tracepoint: Fix overly long tracepoint names x86/perf, static_call: Optimize x86_pmu methods tracepoint: Optimize using static_call() static_call: Allow early init static_call: Add some validation static_call: Handle tail-calls static_call: Add static_call_cond() x86/alternatives: Teach text_poke_bp() to emulate RET static_call: Add simple self-test for static calls x86/static_call: Add inline static call implementation for x86-64 x86/static_call: Add out-of-line static call implementation static_call: Avoid kprobes on inline static_call()s static_call: Add inline static call infrastructure static_call: Add basic static call infrastructure compiler.h: Make __ADDRESSABLE() symbol truly unique jump_label,module: Fix module lifetime for __jump_label_mod_text_reserved() module: Properly propagate MODULE_STATE_COMING failure module: Fix up module_notifier return values ... |
||
---|---|---|
.. | ||
arch/x86 | ||
Documentation | ||
.gitignore | ||
arch.h | ||
Build | ||
builtin-check.c | ||
builtin-orc.c | ||
builtin.h | ||
cfi.h | ||
check.c | ||
check.h | ||
elf.c | ||
elf.h | ||
Makefile | ||
objtool.c | ||
objtool.h | ||
orc_dump.c | ||
orc_gen.c | ||
special.c | ||
special.h | ||
sync-check.sh | ||
warn.h | ||
weak.c |