diff options
author | Linus Torvalds <torvalds@linux-foundation.org> | 2020-06-28 09:42:47 -0700 |
---|---|---|
committer | Linus Torvalds <torvalds@linux-foundation.org> | 2020-06-28 09:42:47 -0700 |
commit | a358505d8af3ab43763e1b49549619528783d2bf (patch) | |
tree | 9486032ab6cfad055f0cffabbf0edc48fce77c52 /Documentation | |
parent | 719fdd32921fb7e3208db8832d32ae1c2d68900f (diff) | |
parent | 2c92d787cc9fad57d05c96bd117782183768258a (diff) | |
download | linux-a358505d8af3ab43763e1b49549619528783d2bf.tar.gz linux-a358505d8af3ab43763e1b49549619528783d2bf.tar.bz2 linux-a358505d8af3ab43763e1b49549619528783d2bf.zip |
Merge tag 'x86_entry_for_5.8' of git://git.kernel.org/pub/scm/linux/kernel/git/tip/tip
Pull x86 entry fixes from Borislav Petkov:
"This is the x86/entry urgent pile which has accumulated since the
merge window.
It is not the smallest but considering the almost complete entry core
rewrite, the amount of fixes to follow is somewhat higher than usual,
which is to be expected.
Peter Zijlstra says:
'These patches address a number of instrumentation issues that were
found after the x86/entry overhaul. When combined with rcu/urgent
and objtool/urgent, these patches make UBSAN/KASAN/KCSAN happy
again.
Part of making this all work is bumping the minimum GCC version for
KASAN builds to gcc-8.3, the reason for this is that the
__no_sanitize_address function attribute is broken in GCC releases
before that.
No known GCC version has a working __no_sanitize_undefined, however
because the only noinstr violation that results from this happens
when an UB is found, we treat it like WARN. That is, we allow it to
violate the noinstr rules in order to get the warning out'"
* tag 'x86_entry_for_5.8' of git://git.kernel.org/pub/scm/linux/kernel/git/tip/tip:
x86/entry: Fix #UD vs WARN more
x86/entry: Increase entry_stack size to a full page
x86/entry: Fixup bad_iret vs noinstr
objtool: Don't consider vmlinux a C-file
kasan: Fix required compiler version
compiler_attributes.h: Support no_sanitize_undefined check with GCC 4
x86/entry, bug: Comment the instrumentation_begin() usage for WARN()
x86/entry, ubsan, objtool: Whitelist __ubsan_handle_*()
x86/entry, cpumask: Provide non-instrumented variant of cpu_is_offline()
compiler_types.h: Add __no_sanitize_{address,undefined} to noinstr
kasan: Bump required compiler version
x86, kcsan: Add __no_kcsan to noinstr
kcsan: Remove __no_kcsan_or_inline
x86, kcsan: Remove __no_kcsan_or_inline usage
Diffstat (limited to 'Documentation')
-rw-r--r-- | Documentation/dev-tools/kcsan.rst | 6 |
1 files changed, 0 insertions, 6 deletions
diff --git a/Documentation/dev-tools/kcsan.rst b/Documentation/dev-tools/kcsan.rst index ce4bbd918648..b38379f06194 100644 --- a/Documentation/dev-tools/kcsan.rst +++ b/Documentation/dev-tools/kcsan.rst @@ -114,12 +114,6 @@ the below options are available: To dynamically limit for which functions to generate reports, see the `DebugFS interface`_ blacklist/whitelist feature. - For ``__always_inline`` functions, replace ``__always_inline`` with - ``__no_kcsan_or_inline`` (which implies ``__always_inline``):: - - static __no_kcsan_or_inline void foo(void) { - ... - * To disable data race detection for a particular compilation unit, add to the ``Makefile``:: |