From mboxrd@z Thu Jan 1 00:00:00 1970 Return-Path: MIME-Version: 1.0 References: <20190820024941.12640-1-dja@axtens.net> <877e6vutiu.fsf@dja-thinkpad.axtens.net> <878sp57z44.fsf@dja-thinkpad.axtens.net> <87a78xgu8o.fsf@dja-thinkpad.axtens.net> <87y2wbf0xx.fsf@dja-thinkpad.axtens.net> In-Reply-To: <87y2wbf0xx.fsf@dja-thinkpad.axtens.net> From: Marco Elver Date: Wed, 20 Nov 2019 09:32:55 +0100 Message-ID: Subject: Re: [PATCH v2 1/2] kasan: support instrumented bitops combined with generic bitops Content-Type: text/plain; charset="UTF-8" Sender: linux-arch-owner@vger.kernel.org List-ID: To: Daniel Axtens Cc: christophe.leroy@c-s.fr, linux-s390@vger.kernel.org, linux-arch , the arch/x86 maintainers , linuxppc-dev@lists.ozlabs.org, kasan-dev On Wed, 20 Nov 2019 at 08:42, Daniel Axtens wrote: > > > But the docs do seem to indicate that it's atomic (for whatever that > > means for a single read operation?), so you are right, it should live in > > instrumented-atomic.h. > > Actually, on further inspection, test_bit has lived in > bitops/non-atomic.h since it was added in 4117b02132d1 ("[PATCH] bitops: > generic __{,test_and_}{set,clear,change}_bit() and test_bit()") > > So to match that, the wrapper should live in instrumented-non-atomic.h > too. > > If test_bit should move, that would need to be a different patch. But I > don't really know if it makes too much sense to stress about a read > operation, as opposed to a read/modify/write... That's fair enough. I suppose this can stay where it is because it's not hurting anyone per-se, but the only bad thing about it is that kernel-api documentation will present test_bit() in non-atomic operations. Thanks, -- Marco From mboxrd@z Thu Jan 1 00:00:00 1970 From: Marco Elver Subject: Re: [PATCH v2 1/2] kasan: support instrumented bitops combined with generic bitops Date: Wed, 20 Nov 2019 09:32:55 +0100 Message-ID: References: <20190820024941.12640-1-dja@axtens.net> <877e6vutiu.fsf@dja-thinkpad.axtens.net> <878sp57z44.fsf@dja-thinkpad.axtens.net> <87a78xgu8o.fsf@dja-thinkpad.axtens.net> <87y2wbf0xx.fsf@dja-thinkpad.axtens.net> Mime-Version: 1.0 Content-Type: text/plain; charset="UTF-8" Return-path: In-Reply-To: <87y2wbf0xx.fsf@dja-thinkpad.axtens.net> List-Unsubscribe: , List-Archive: List-Post: List-Help: List-Subscribe: , Errors-To: linuxppc-dev-bounces+glppe-linuxppc-embedded-2=m.gmane.org@lists.ozlabs.org Sender: "Linuxppc-dev" To: Daniel Axtens Cc: linux-s390@vger.kernel.org, the arch/x86 maintainers , kasan-dev , linux-arch , linuxppc-dev@lists.ozlabs.org List-Id: linux-arch.vger.kernel.org On Wed, 20 Nov 2019 at 08:42, Daniel Axtens wrote: > > > But the docs do seem to indicate that it's atomic (for whatever that > > means for a single read operation?), so you are right, it should live in > > instrumented-atomic.h. > > Actually, on further inspection, test_bit has lived in > bitops/non-atomic.h since it was added in 4117b02132d1 ("[PATCH] bitops: > generic __{,test_and_}{set,clear,change}_bit() and test_bit()") > > So to match that, the wrapper should live in instrumented-non-atomic.h > too. > > If test_bit should move, that would need to be a different patch. But I > don't really know if it makes too much sense to stress about a read > operation, as opposed to a read/modify/write... That's fair enough. I suppose this can stay where it is because it's not hurting anyone per-se, but the only bad thing about it is that kernel-api documentation will present test_bit() in non-atomic operations. Thanks, -- Marco