From owner-dev-commits-src-main@freebsd.org Wed Mar 3 14:55:42 2021 Return-Path: Delivered-To: dev-commits-src-main@mailman.nyi.freebsd.org Received: from mx1.freebsd.org (mx1.freebsd.org [IPv6:2610:1c1:1:606c::19:1]) by mailman.nyi.freebsd.org (Postfix) with ESMTP id 0813E56CB4A; Wed, 3 Mar 2021 14:55:42 +0000 (UTC) (envelope-from andrew@freebsd.org) Received: from fry.fubar.geek.nz (fry.fubar.geek.nz [139.59.165.16]) by mx1.freebsd.org (Postfix) with ESMTP id 4DrH9n4tGTz3s4w; Wed, 3 Mar 2021 14:55:41 +0000 (UTC) (envelope-from andrew@freebsd.org) Received: from [192.168.42.21] (cpc91232-cmbg18-2-0-cust554.5-4.cable.virginm.net [82.2.126.43]) by fry.fubar.geek.nz (Postfix) with ESMTPSA id 9326D4E71F; Wed, 3 Mar 2021 14:55:34 +0000 (UTC) From: Andrew Turner Message-Id: Mime-Version: 1.0 (Mac OS X Mail 12.4 \(3445.104.17\)) Subject: Re: git: 28d945204ea1 - main - Handle functions that use a nop in the arm64 fbt Date: Wed, 3 Mar 2021 14:55:33 +0000 In-Reply-To: Cc: "src-committers@freebsd.org" , "dev-commits-src-all@freebsd.org" , "dev-commits-src-main@freebsd.org" To: Jessica Clarke References: <202103031426.123EQoM5082920@gitrepo.freebsd.org> X-Mailer: Apple Mail (2.3445.104.17) X-Rspamd-Queue-Id: 4DrH9n4tGTz3s4w X-Spamd-Bar: ---- Authentication-Results: mx1.freebsd.org; none X-Spamd-Result: default: False [-4.00 / 15.00]; REPLY(-4.00)[] Content-Type: text/plain; charset=utf-8 Content-Transfer-Encoding: quoted-printable X-Content-Filtered-By: Mailman/MimeDel 2.1.34 X-BeenThere: dev-commits-src-main@freebsd.org X-Mailman-Version: 2.1.34 Precedence: list List-Id: Commit messages for the main branch of the src repository List-Unsubscribe: , List-Archive: List-Post: List-Help: List-Subscribe: , X-List-Received-Date: Wed, 03 Mar 2021 14:55:42 -0000 > On 3 Mar 2021, at 14:37, Jessica Clarke wrote: >=20 > On 3 Mar 2021, at 14:29, Jessica Clarke > wrote: >> On 3 Mar 2021, at 14:26, Andrew Turner wrote: >>>=20 >>> The branch main has been updated by andrew: >>>=20 >>> URL: = https://cgit.FreeBSD.org/src/commit/?id=3D28d945204ea1014d7de6906af8470ed8= b3311335 >>>=20 >>> commit 28d945204ea1014d7de6906af8470ed8b3311335 >>> Author: Andrew Turner >>> AuthorDate: 2021-01-13 11:08:19 +0000 >>> Commit: Andrew Turner >>> CommitDate: 2021-03-03 14:18:03 +0000 >>>=20 >>> Handle functions that use a nop in the arm64 fbt >>>=20 >>> To trace leaf asm functions we can insert a single nop instruction = as >>> the first instruction in a function and trigger off this. >>>=20 >>> Reviewed by: gnn >>> Sponsored by: Innovate UK >>> Differential Revision: https://reviews.freebsd.org/D28132 >>> --- >>> sys/arm64/include/asm.h | 8 +++- >>> .../contrib/opensolaris/uts/common/sys/dtrace.h | 2 + >>> sys/cddl/dev/dtrace/aarch64/dtrace_subr.c | 5 +++ >>> sys/cddl/dev/fbt/aarch64/fbt_isa.c | 51 = ++++++++++++++-------- >>> 4 files changed, 46 insertions(+), 20 deletions(-) >>>=20 >>> diff --git a/sys/arm64/include/asm.h b/sys/arm64/include/asm.h >>> index 05e618500e59..32b79d256e80 100644 >>> --- a/sys/arm64/include/asm.h >>> +++ b/sys/arm64/include/asm.h >>> @@ -38,9 +38,15 @@ >>>=20 >>> #define _C_LABEL(x) x >>>=20 >>> +#ifdef KDTRACE_HOOKS >>> +#define DTRACE_NOP nop >>> +#else >>> +#define DTRACE_NOP >>> +#endif >>> + >>> #define LENTRY(sym) = \ >>> .text; .align 2; .type sym,#function; sym: \ >>> - .cfi_startproc >>> + .cfi_startproc; DTRACE_NOP >>> #define ENTRY(sym) = \ >>> .globl sym; LENTRY(sym) >>=20 >> Doesn't this mean ENTRY incorrectly also has the nop? >=20 > Hm, right, the L in LENTRY means local not leaf. Isn't this a problem > though? (L)ENTRY are perfectly legal to use for non-leaf assembly > functions today. Shouldn't there be separate ones specifically for = leaf > functions if you want to treat them differently? Other than early boot handling, pmap_switch, and the exception handlers = I think we only have a few non-leaf asm functions on arm64. The only = ones I can think of use tail recursion, e.g. memmove -> memcpy when = possible. Other than exception handlers these functions don=E2=80=99t = have the needed instructions to manage the stack frame as they don=E2=80=99= t use any stack space. I decided it was easier to add the nop = instruction to the start of function than try to create an unneeded = stack frame. Andrew=