From mboxrd@z Thu Jan 1 00:00:00 1970 From: samitolvanen@google.com (Sami Tolvanen) Date: Fri, 17 Nov 2017 10:49:55 -0800 Subject: [PATCH v2 08/18] arm64: don't disable ADR_PREL_PG_HI21* with ARM64_ERRATUM_843419 In-Reply-To: References: <20171116114406.GB9361@arm.com> <20171116163132.GC94341@samitolvanen.mtv.corp.google.com> <20171116213702.GA10616@samitolvanen.mtv.corp.google.com> <20171116230939.GA38360@samitolvanen.mtv.corp.google.com> <20171116235019.GA55530@samitolvanen.mtv.corp.google.com> Message-ID: <20171117184955.GA39391@samitolvanen.mtv.corp.google.com> To: linux-arm-kernel@lists.infradead.org List-Id: linux-arm-kernel.lists.infradead.org On Fri, Nov 17, 2017 at 09:54:48AM +0000, Ard Biesheuvel wrote: > OK, so all IR objects are converted into a single .o file > encapsulating the module image. Does this give the same benefits as > LTO linking IR objects to a fully linked executable? Yes, it does. > Even if it does, partial linking is not something the toolchain > people are usually crazy about, so it would be nice to have some > confirmation that this is a usage model that is fully supported. I confirmed with our LLVM developers that while this is less common, it's fully supported. I will also drop this patch in v3 as passing code model to LLVMgold fixes the issue with LTO. Sami