From mboxrd@z Thu Jan 1 00:00:00 1970 From: will.deacon@arm.com (Will Deacon) Date: Wed, 18 Mar 2015 10:39:02 +0000 Subject: [PATCH 2/2] ARM: psci: move psci firmware calls out of line In-Reply-To: <20150318103556.GD8656@n2100.arm.linux.org.uk> References: <1424866236-12482-1-git-send-email-will.deacon@arm.com> <1424866236-12482-2-git-send-email-will.deacon@arm.com> <20150318102029.GC8656@n2100.arm.linux.org.uk> <20150318103014.GD18951@arm.com> <20150318103556.GD8656@n2100.arm.linux.org.uk> Message-ID: <20150318103902.GF18951@arm.com> To: linux-arm-kernel@lists.infradead.org List-Id: linux-arm-kernel.lists.infradead.org On Wed, Mar 18, 2015 at 10:35:57AM +0000, Russell King - ARM Linux wrote: > On Wed, Mar 18, 2015 at 10:30:14AM +0000, Will Deacon wrote: > > On Wed, Mar 18, 2015 at 10:20:29AM +0000, Russell King - ARM Linux wrote: > > > Why not convert these to: > > > > > > static int __naked __invoke_psci_fn_hvc(u32 function_id, u32 arg0, u32 arg1, > > > u32 arg2) > > > { > > > asm( > > > __HVC(0) > > > "bx lr"); > > > } > > > > > > ? > > > > I tried this, but the compiler printed a diagnostic along the lines of > > "Ignoring __naked attribute", so we moved the functions out-of-line like > > we have on arm64 instead. Assuming it worked reliably, what's the > > advantage of __naked over having these out-of-line? > > Note that the above isn't using any asm constraints/arguments. The GCC > manual states what can be safely included: > > `naked' > Use this attribute on the ARM, AVR, IP2K, RX and SPU ports to > indicate that the specified function does not need > prologue/epilogue sequences generated by the compiler. It is up > to the programmer to provide these sequences. The only statements > > that can be safely included in naked functions are `asm' > > statements that do not have operands. All other statements, > ^^^^^^^^^^^^^^^^^^^^^^^^^^ > > including declarations of local variables, `if' statements, and so > forth, should be avoided. Naked functions should be used to > implement the body of an assembly function, while allowing the > compiler to construct the requisite function declaration for the > assembler. > > The advantage is that you have locality of code, and we don't need to > spring up lots of assembly files along side their .c files. > > I'd guess that if you try to declare local variables etc, the compiler > will just ignore the naked attribute, because it then can't give the > guarantees that the function will be truely naked. In which case, I can reinvestigate when I get some spare cycles. IIRC, the problem only cropped up with allmodconfig, because we compile with -pg but it's all a bit hazy... Will