From nobody Sat Feb 7 15:16:54 2026 Return-Path: X-Spam-Checker-Version: SpamAssassin 3.4.0 (2014-02-07) on aws-us-west-2-korg-lkml-1.web.codeaurora.org Received: from vger.kernel.org (vger.kernel.org [23.128.96.18]) by smtp.lore.kernel.org (Postfix) with ESMTP id 0C0A5ECAAA1 for ; Fri, 28 Oct 2022 16:48:04 +0000 (UTC) Received: (majordomo@vger.kernel.org) by vger.kernel.org via listexpand id S229994AbiJ1QsC (ORCPT ); Fri, 28 Oct 2022 12:48:02 -0400 Received: from lindbergh.monkeyblade.net ([23.128.96.19]:39070 "EHLO lindbergh.monkeyblade.net" rhost-flags-OK-OK-OK-OK) by vger.kernel.org with ESMTP id S229482AbiJ1Qr6 (ORCPT ); Fri, 28 Oct 2022 12:47:58 -0400 Received: from pandora.armlinux.org.uk (pandora.armlinux.org.uk [IPv6:2001:4d48:ad52:32c8:5054:ff:fe00:142]) by lindbergh.monkeyblade.net (Postfix) with ESMTPS id DD5461EEA3F for ; Fri, 28 Oct 2022 09:47:56 -0700 (PDT) DKIM-Signature: v=1; a=rsa-sha256; q=dns/txt; c=relaxed/relaxed; d=armlinux.org.uk; s=pandora-2019; h=Date:Sender:Message-Id:Content-Type: Content-Transfer-Encoding:MIME-Version:Subject:Cc:To:From:References: In-Reply-To:Reply-To:Content-ID:Content-Description:Resent-Date:Resent-From: Resent-Sender:Resent-To:Resent-Cc:Resent-Message-ID:List-Id:List-Help: List-Unsubscribe:List-Subscribe:List-Post:List-Owner:List-Archive; bh=QKytlREYiTLpgxfFh35rP21mKKtH4zCnPdMR02CiPB4=; b=tUYLUTBO4ITtmpwmX0vst1lWZZ DVnidmNbD/Q7dg/D0oMx6J93/GV4M4yHYoR8tkpq7920ihF9F+AkLTW5rjFNS9hWIJIt0zPwRzfWJ ufzAIXuYC6yl7KKuRcbvKLwSIW3F9p/cYgZl47J4oYVDBp5pFnr61O+9o6Ap7kxIVrjhHH2TF+rVa t0PTvqd37R1IF3iH4iWuFiptMNoT7phGxK81N6M1dmUp5Hujz+k8q+9lxX5lfLBBfDLHtVuGDmAML gm2Guc6v0yrxAtCMba0DDGai43H5vWv6isL/87B3MwJfJBr4zwuxDPh+czo2GVRqa6DDwxmWyT+Z/ LAX4cMqg==; Received: from e0022681537dd.dyn.armlinux.org.uk ([fd8f:7570:feb6:1:222:68ff:fe15:37dd]:55688 helo=rmk-PC.armlinux.org.uk) by pandora.armlinux.org.uk with esmtpsa (TLS1.3) tls TLS_ECDHE_RSA_WITH_AES_256_GCM_SHA384 (Exim 4.94.2) (envelope-from ) id 1ooSWP-00006W-OE; Fri, 28 Oct 2022 17:47:53 +0100 Received: from rmk by rmk-PC.armlinux.org.uk with local (Exim 4.94.2) (envelope-from ) id 1ooSWP-000FDy-5t; Fri, 28 Oct 2022 17:47:53 +0100 In-Reply-To: References: From: "Russell King (Oracle)" To: Yury Norov Cc: Catalin Marinas , Mark Rutland , Will Deacon , Linux Kernel Mailing List , Linus Torvalds , linux-arm-kernel@lists.infradead.org Subject: [PATCH 1/5] ARM: findbit: document ARMv5 bit offset calculation MIME-Version: 1.0 Content-Disposition: inline Content-Transfer-Encoding: quoted-printable Content-Type: text/plain; charset="utf-8" Message-Id: Sender: Russell King Date: Fri, 28 Oct 2022 17:47:53 +0100 Precedence: bulk List-ID: X-Mailing-List: linux-kernel@vger.kernel.org Document the ARMv5 bit offset calculation code. Signed-off-by: Russell King (Oracle) --- arch/arm/lib/findbit.S | 8 ++++---- 1 file changed, 4 insertions(+), 4 deletions(-) diff --git a/arch/arm/lib/findbit.S b/arch/arm/lib/findbit.S index 7fd3600db8ef..4c584bc4704b 100644 --- a/arch/arm/lib/findbit.S +++ b/arch/arm/lib/findbit.S @@ -172,10 +172,10 @@ ENDPROC(_find_next_bit_be) .L_found: #if __LINUX_ARM_ARCH__ >=3D 5 rsb r0, r3, #0 - and r3, r3, r0 - clz r3, r3 - rsb r3, r3, #31 - add r0, r2, r3 + and r3, r3, r0 @ mask out lowest bit set + clz r3, r3 @ count high zero bits + rsb r3, r3, #31 @ offset of first set bit + add r0, r2, r3 @ add offset of first set bit #else tst r3, #0x0f addeq r2, r2, #4 --=20 2.30.2 From nobody Sat Feb 7 15:16:54 2026 Return-Path: X-Spam-Checker-Version: SpamAssassin 3.4.0 (2014-02-07) on aws-us-west-2-korg-lkml-1.web.codeaurora.org Received: from vger.kernel.org (vger.kernel.org [23.128.96.18]) by smtp.lore.kernel.org (Postfix) with ESMTP id EB65AFA3742 for ; Fri, 28 Oct 2022 16:48:08 +0000 (UTC) Received: (majordomo@vger.kernel.org) by vger.kernel.org via listexpand id S230036AbiJ1QsG (ORCPT ); Fri, 28 Oct 2022 12:48:06 -0400 Received: from lindbergh.monkeyblade.net ([23.128.96.19]:39108 "EHLO lindbergh.monkeyblade.net" rhost-flags-OK-OK-OK-OK) by vger.kernel.org with ESMTP id S229988AbiJ1QsB (ORCPT ); Fri, 28 Oct 2022 12:48:01 -0400 Received: from pandora.armlinux.org.uk (pandora.armlinux.org.uk [IPv6:2001:4d48:ad52:32c8:5054:ff:fe00:142]) by lindbergh.monkeyblade.net (Postfix) with ESMTPS id E0E141EEF01 for ; Fri, 28 Oct 2022 09:48:00 -0700 (PDT) DKIM-Signature: v=1; a=rsa-sha256; q=dns/txt; c=relaxed/relaxed; d=armlinux.org.uk; s=pandora-2019; h=Date:Sender:Message-Id:Content-Type: Content-Transfer-Encoding:MIME-Version:Subject:Cc:To:From:References: In-Reply-To:Reply-To:Content-ID:Content-Description:Resent-Date:Resent-From: Resent-Sender:Resent-To:Resent-Cc:Resent-Message-ID:List-Id:List-Help: List-Unsubscribe:List-Subscribe:List-Post:List-Owner:List-Archive; bh=noL9/NeDUTGkTM9TkwKb+qB4g92MFSxxMDGz/3I26JY=; b=tTloeFQWZ0uoDO5GJamaoriva2 cIFPvVdZJsD4tqdlzlbICcPKZ7oX/ZY+D14hvxMfOE+WoHtE30EY66VVrh/WS/z13sIo3uWCMbyhm QFC5lsr6yq8M4lpXjTYjcE3JU8CqnfGeT5tHIKgpxxDcVg2Ghhh3Tym6zM4wrXQcWLclX501T6R3a mTs50+ZSY4ziba5+FuQlDKf2TEPIf9Ma+dnc18tsVY8VcD2LA+scbJScXjkglrg8bvjKMuh6tkJlc 3+QIcPkgxg85h+O+Pp1gzgO67hoiR1tQ1WjteSZI3bKACJ8dzRr1Q5+RLw2WitLDyWA4d7OGa9Dn7 bHS7g84Q==; Received: from e0022681537dd.dyn.armlinux.org.uk ([fd8f:7570:feb6:1:222:68ff:fe15:37dd]:55696 helo=rmk-PC.armlinux.org.uk) by pandora.armlinux.org.uk with esmtpsa (TLS1.3) tls TLS_ECDHE_RSA_WITH_AES_256_GCM_SHA384 (Exim 4.94.2) (envelope-from ) id 1ooSWU-00006h-RW; Fri, 28 Oct 2022 17:47:58 +0100 Received: from rmk by rmk-PC.armlinux.org.uk with local (Exim 4.94.2) (envelope-from ) id 1ooSWU-000FE4-9D; Fri, 28 Oct 2022 17:47:58 +0100 In-Reply-To: References: From: "Russell King (Oracle)" To: Yury Norov Cc: Catalin Marinas , Mark Rutland , Will Deacon , Linux Kernel Mailing List , Linus Torvalds , linux-arm-kernel@lists.infradead.org Subject: [PATCH 2/5] ARM: findbit: provide more efficient ARMv7 implementation MIME-Version: 1.0 Content-Disposition: inline Content-Transfer-Encoding: quoted-printable Content-Type: text/plain; charset="utf-8" Message-Id: Sender: Russell King Date: Fri, 28 Oct 2022 17:47:58 +0100 Precedence: bulk List-ID: X-Mailing-List: linux-kernel@vger.kernel.org Provide a more efficient ARMv7 implementation to determine the first set bit in the supplied value. Signed-off-by: Russell King (Oracle) --- arch/arm/lib/findbit.S | 6 +++++- 1 file changed, 5 insertions(+), 1 deletion(-) diff --git a/arch/arm/lib/findbit.S b/arch/arm/lib/findbit.S index 4c584bc4704b..256e095d490b 100644 --- a/arch/arm/lib/findbit.S +++ b/arch/arm/lib/findbit.S @@ -170,7 +170,11 @@ ENDPROC(_find_next_bit_be) * One or more bits in the LSB of r3 are assumed to be set. */ .L_found: -#if __LINUX_ARM_ARCH__ >=3D 5 +#if __LINUX_ARM_ARCH__ >=3D 7 + rbit r3, r3 @ reverse bits + clz r3, r3 @ count high zero bits + add r0, r2, r3 @ add offset of first set bit +#elif __LINUX_ARM_ARCH__ >=3D 5 rsb r0, r3, #0 and r3, r3, r0 @ mask out lowest bit set clz r3, r3 @ count high zero bits --=20 2.30.2 From nobody Sat Feb 7 15:16:54 2026 Return-Path: X-Spam-Checker-Version: SpamAssassin 3.4.0 (2014-02-07) on aws-us-west-2-korg-lkml-1.web.codeaurora.org Received: from vger.kernel.org (vger.kernel.org [23.128.96.18]) by smtp.lore.kernel.org (Postfix) with ESMTP id 46AA8ECAAA1 for ; Fri, 28 Oct 2022 16:48:23 +0000 (UTC) Received: (majordomo@vger.kernel.org) by vger.kernel.org via listexpand id S230171AbiJ1QsU (ORCPT ); Fri, 28 Oct 2022 12:48:20 -0400 Received: from lindbergh.monkeyblade.net ([23.128.96.19]:39384 "EHLO lindbergh.monkeyblade.net" rhost-flags-OK-OK-OK-OK) by vger.kernel.org with ESMTP id S230039AbiJ1QsI (ORCPT ); Fri, 28 Oct 2022 12:48:08 -0400 Received: from pandora.armlinux.org.uk (pandora.armlinux.org.uk [IPv6:2001:4d48:ad52:32c8:5054:ff:fe00:142]) by lindbergh.monkeyblade.net (Postfix) with ESMTPS id 294F120DB74 for ; Fri, 28 Oct 2022 09:48:06 -0700 (PDT) DKIM-Signature: v=1; a=rsa-sha256; q=dns/txt; c=relaxed/relaxed; d=armlinux.org.uk; s=pandora-2019; h=Date:Sender:Message-Id:Content-Type: Content-Transfer-Encoding:MIME-Version:Subject:Cc:To:From:References: In-Reply-To:Reply-To:Content-ID:Content-Description:Resent-Date:Resent-From: Resent-Sender:Resent-To:Resent-Cc:Resent-Message-ID:List-Id:List-Help: List-Unsubscribe:List-Subscribe:List-Post:List-Owner:List-Archive; bh=WkDjf+3IQgCyMglzhEzUvxitZ2xNrXhv5I3AR6iDziY=; b=M/Gbjsk9bvCVCw0Chxu+P8YgKs GaYtSL13ilKICTNDPb8tJ4/atXRKoeph3SLC8sixyssIPvIoQcXbbHQBdHo9xHp2tuikvozlov+QS U4g9MXMyB6nFkk6GRR7Sy4LQAswi9CxnUr+cwup5sG4I1qwIbjrd8y9nA5xcAXk4TZ4Zq5jCZoW4A Wz9uHo+PTHBsoXU2wnS8pkctAY1FZUYMdsErxsA8sn4NYeN8NaPsMxQCRgM5AVmHmhbuZGzfEpDaV AaseJ2fuTdmFPu1WTp5Z+WDeF5Bp26fgtkJGzTsl4R5/KQD7WQXJihwKRmvQOpg5tcnAgtAChjsu6 z9qUjezw==; Received: from e0022681537dd.dyn.armlinux.org.uk ([fd8f:7570:feb6:1:222:68ff:fe15:37dd]:33780 helo=rmk-PC.armlinux.org.uk) by pandora.armlinux.org.uk with esmtpsa (TLS1.3) tls TLS_ECDHE_RSA_WITH_AES_256_GCM_SHA384 (Exim 4.94.2) (envelope-from ) id 1ooSWZ-00006s-W3; Fri, 28 Oct 2022 17:48:04 +0100 Received: from rmk by rmk-PC.armlinux.org.uk with local (Exim 4.94.2) (envelope-from ) id 1ooSWZ-000FEA-CU; Fri, 28 Oct 2022 17:48:03 +0100 In-Reply-To: References: From: "Russell King (Oracle)" To: Yury Norov Cc: Catalin Marinas , Mark Rutland , Will Deacon , Linux Kernel Mailing List , Linus Torvalds , linux-arm-kernel@lists.infradead.org Subject: [PATCH 3/5] ARM: findbit: convert to macros MIME-Version: 1.0 Content-Disposition: inline Content-Transfer-Encoding: quoted-printable Content-Type: text/plain; charset="utf-8" Message-Id: Sender: Russell King Date: Fri, 28 Oct 2022 17:48:03 +0100 Precedence: bulk List-ID: X-Mailing-List: linux-kernel@vger.kernel.org Since the pairs of _find_first and _find_next functions are pretty similar, use macros to generate this code. This commit does not change the generated code. Signed-off-by: Russell King (Oracle) --- arch/arm/lib/findbit.S | 158 +++++++++++------------------------------ 1 file changed, 42 insertions(+), 116 deletions(-) diff --git a/arch/arm/lib/findbit.S b/arch/arm/lib/findbit.S index 256e095d490b..8280f66d38a5 100644 --- a/arch/arm/lib/findbit.S +++ b/arch/arm/lib/findbit.S @@ -14,155 +14,81 @@ #include .text =20 -/* - * Purpose : Find a 'zero' bit - * Prototype: int find_first_zero_bit(void *addr, unsigned int maxbit); - */ -ENTRY(_find_first_zero_bit_le) - teq r1, #0=09 + .macro find_first, endian, set, name +ENTRY(_find_first_\name\()bit_\endian) + teq r1, #0 beq 3f mov r2, #0 1: + .ifc \endian, be + eor r3, r2, #0x18 + ARM( ldrb r3, [r0, r3, lsr #3] ) + THUMB( lsr r3, #3 ) + THUMB( ldrb r3, [r0, r3] ) + .else ARM( ldrb r3, [r0, r2, lsr #3] ) THUMB( lsr r3, r2, #3 ) THUMB( ldrb r3, [r0, r3] ) + .endif + .ifeq \set eors r3, r3, #0xff @ invert bits + .else + movs r3, r3 + .endif bne .L_found @ any now set - found zero bit add r2, r2, #8 @ next bit pointer 2: cmp r2, r1 @ any more? blo 1b 3: mov r0, r1 @ no free bits ret lr -ENDPROC(_find_first_zero_bit_le) +ENDPROC(_find_first_\name\()bit_\endian) + .endm =20 -/* - * Purpose : Find next 'zero' bit - * Prototype: int find_next_zero_bit(void *addr, unsigned int maxbit, int = offset) - */ -ENTRY(_find_next_zero_bit_le) + .macro find_next, endian, set, name +ENTRY(_find_next_\name\()bit_\endian) cmp r2, r1 bhs 3b ands ip, r2, #7 beq 1b @ If new byte, goto old routine + .ifc \endian, be + eor r3, r2, #0x18 + ARM( ldrb r3, [r0, r3, lsr #3] ) + THUMB( lsr r3, #3 ) + THUMB( ldrb r3, [r0, r3] ) + .else ARM( ldrb r3, [r0, r2, lsr #3] ) THUMB( lsr r3, r2, #3 ) THUMB( ldrb r3, [r0, r3] ) + .endif + .ifeq \set eor r3, r3, #0xff @ now looking for a 1 bit + .endif movs r3, r3, lsr ip @ shift off unused bits bne .L_found orr r2, r2, #7 @ if zero, then no bits here add r2, r2, #1 @ align bit pointer b 2b @ loop for next bit -ENDPROC(_find_next_zero_bit_le) +ENDPROC(_find_next_\name\()bit_\endian) + .endm =20 -/* - * Purpose : Find a 'one' bit - * Prototype: int find_first_bit(const unsigned long *addr, unsigned int m= axbit); - */ -ENTRY(_find_first_bit_le) - teq r1, #0=09 - beq 3f - mov r2, #0 -1: - ARM( ldrb r3, [r0, r2, lsr #3] ) - THUMB( lsr r3, r2, #3 ) - THUMB( ldrb r3, [r0, r3] ) - movs r3, r3 - bne .L_found @ any now set - found zero bit - add r2, r2, #8 @ next bit pointer -2: cmp r2, r1 @ any more? - blo 1b -3: mov r0, r1 @ no free bits - ret lr -ENDPROC(_find_first_bit_le) + .macro find_bit, endian, set, name + find_first \endian, \set, \name + find_next \endian, \set, \name + .endm =20 -/* - * Purpose : Find next 'one' bit - * Prototype: int find_next_zero_bit(void *addr, unsigned int maxbit, int = offset) - */ -ENTRY(_find_next_bit_le) - cmp r2, r1 - bhs 3b - ands ip, r2, #7 - beq 1b @ If new byte, goto old routine - ARM( ldrb r3, [r0, r2, lsr #3] ) - THUMB( lsr r3, r2, #3 ) - THUMB( ldrb r3, [r0, r3] ) - movs r3, r3, lsr ip @ shift off unused bits - bne .L_found - orr r2, r2, #7 @ if zero, then no bits here - add r2, r2, #1 @ align bit pointer - b 2b @ loop for next bit -ENDPROC(_find_next_bit_le) +/* _find_first_zero_bit_le and _find_next_zero_bit_le */ + find_bit le, 0, zero_ =20 -#ifdef __ARMEB__ +/* _find_first_bit_le and _find_next_bit_le */ + find_bit le, 1 =20 -ENTRY(_find_first_zero_bit_be) - teq r1, #0 - beq 3f - mov r2, #0 -1: eor r3, r2, #0x18 @ big endian byte ordering - ARM( ldrb r3, [r0, r3, lsr #3] ) - THUMB( lsr r3, #3 ) - THUMB( ldrb r3, [r0, r3] ) - eors r3, r3, #0xff @ invert bits - bne .L_found @ any now set - found zero bit - add r2, r2, #8 @ next bit pointer -2: cmp r2, r1 @ any more? - blo 1b -3: mov r0, r1 @ no free bits - ret lr -ENDPROC(_find_first_zero_bit_be) +#ifdef __ARMEB__ =20 -ENTRY(_find_next_zero_bit_be) - cmp r2, r1 - bhs 3b - ands ip, r2, #7 - beq 1b @ If new byte, goto old routine - eor r3, r2, #0x18 @ big endian byte ordering - ARM( ldrb r3, [r0, r3, lsr #3] ) - THUMB( lsr r3, #3 ) - THUMB( ldrb r3, [r0, r3] ) - eor r3, r3, #0xff @ now looking for a 1 bit - movs r3, r3, lsr ip @ shift off unused bits - bne .L_found - orr r2, r2, #7 @ if zero, then no bits here - add r2, r2, #1 @ align bit pointer - b 2b @ loop for next bit -ENDPROC(_find_next_zero_bit_be) +/* _find_first_zero_bit_be and _find_next_zero_bit_be */ + find_bit be, 0, zero_ =20 -ENTRY(_find_first_bit_be) - teq r1, #0 - beq 3f - mov r2, #0 -1: eor r3, r2, #0x18 @ big endian byte ordering - ARM( ldrb r3, [r0, r3, lsr #3] ) - THUMB( lsr r3, #3 ) - THUMB( ldrb r3, [r0, r3] ) - movs r3, r3 - bne .L_found @ any now set - found zero bit - add r2, r2, #8 @ next bit pointer -2: cmp r2, r1 @ any more? - blo 1b -3: mov r0, r1 @ no free bits - ret lr -ENDPROC(_find_first_bit_be) - -ENTRY(_find_next_bit_be) - cmp r2, r1 - bhs 3b - ands ip, r2, #7 - beq 1b @ If new byte, goto old routine - eor r3, r2, #0x18 @ big endian byte ordering - ARM( ldrb r3, [r0, r3, lsr #3] ) - THUMB( lsr r3, #3 ) - THUMB( ldrb r3, [r0, r3] ) - movs r3, r3, lsr ip @ shift off unused bits - bne .L_found - orr r2, r2, #7 @ if zero, then no bits here - add r2, r2, #1 @ align bit pointer - b 2b @ loop for next bit -ENDPROC(_find_next_bit_be) +/* _find_first_bit_be and _find_next_bit_be */ + find_bit be, 1 =20 #endif =20 --=20 2.30.2 From nobody Sat Feb 7 15:16:54 2026 Return-Path: X-Spam-Checker-Version: SpamAssassin 3.4.0 (2014-02-07) on aws-us-west-2-korg-lkml-1.web.codeaurora.org Received: from vger.kernel.org (vger.kernel.org [23.128.96.18]) by smtp.lore.kernel.org (Postfix) with ESMTP id CACD2ECAAA1 for ; Fri, 28 Oct 2022 16:48:35 +0000 (UTC) Received: (majordomo@vger.kernel.org) by vger.kernel.org via listexpand id S230231AbiJ1Qse (ORCPT ); Fri, 28 Oct 2022 12:48:34 -0400 Received: from lindbergh.monkeyblade.net ([23.128.96.19]:39660 "EHLO lindbergh.monkeyblade.net" rhost-flags-OK-OK-OK-OK) by vger.kernel.org with ESMTP id S230147AbiJ1QsO (ORCPT ); Fri, 28 Oct 2022 12:48:14 -0400 Received: from pandora.armlinux.org.uk (pandora.armlinux.org.uk [IPv6:2001:4d48:ad52:32c8:5054:ff:fe00:142]) by lindbergh.monkeyblade.net (Postfix) with ESMTPS id 75FBB20FB25 for ; Fri, 28 Oct 2022 09:48:11 -0700 (PDT) DKIM-Signature: v=1; a=rsa-sha256; q=dns/txt; c=relaxed/relaxed; d=armlinux.org.uk; s=pandora-2019; h=Date:Sender:Message-Id:Content-Type: Content-Transfer-Encoding:MIME-Version:Subject:Cc:To:From:References: In-Reply-To:Reply-To:Content-ID:Content-Description:Resent-Date:Resent-From: Resent-Sender:Resent-To:Resent-Cc:Resent-Message-ID:List-Id:List-Help: List-Unsubscribe:List-Subscribe:List-Post:List-Owner:List-Archive; bh=awLfhDYQKzr/5OLyGPM7CZguAgBNbZFtXaI32A3EvdI=; b=CO64jMczJw9bQw0Y3YstvDrriM M+W8fedhsaHxOfUSNWbJRsCBF9rAcxapIXf0Q+k955dpSOJDszQvU5HVXUWsWrkqzWJPK/g+/1k1i Z/3VObBwbMlz4Lj+7cCoKw3nsxnl7kWr15DLtET6C/fi3XOKJMgvZy5cHD1dz6pPzhukAgG/TZBMO x8EinmXLmk2YovZTrFKXEMB1CwzP1DSWAKLqosYnUJ4EBn10Efc57gXrVpn2e2HqO12vEjSyeu3zt Gfl/JjOzUvA9f27nbL2i0gtTAz5259q+dubis4/J1EQZQqAkFhxMBc7SgruCIV5YHCukHdTjHpZAJ BB0OGYoA==; Received: from e0022681537dd.dyn.armlinux.org.uk ([fd8f:7570:feb6:1:222:68ff:fe15:37dd]:33794 helo=rmk-PC.armlinux.org.uk) by pandora.armlinux.org.uk with esmtpsa (TLS1.3) tls TLS_ECDHE_RSA_WITH_AES_256_GCM_SHA384 (Exim 4.94.2) (envelope-from ) id 1ooSWf-000073-2Y; Fri, 28 Oct 2022 17:48:09 +0100 Received: from rmk by rmk-PC.armlinux.org.uk with local (Exim 4.94.2) (envelope-from ) id 1ooSWe-000FEG-G7; Fri, 28 Oct 2022 17:48:08 +0100 In-Reply-To: References: From: "Russell King (Oracle)" To: Yury Norov Cc: Catalin Marinas , Mark Rutland , Will Deacon , Linux Kernel Mailing List , Linus Torvalds , linux-arm-kernel@lists.infradead.org Subject: [PATCH 4/5] ARM: findbit: operate by words MIME-Version: 1.0 Content-Disposition: inline Content-Transfer-Encoding: quoted-printable Content-Type: text/plain; charset="utf-8" Message-Id: Sender: Russell King Date: Fri, 28 Oct 2022 17:48:08 +0100 Precedence: bulk List-ID: X-Mailing-List: linux-kernel@vger.kernel.org Convert the implementations to operate on words rather than bytes which makes bitmap searching faster. Signed-off-by: Russell King (Oracle) --- arch/arm/include/asm/assembler.h | 6 +++ arch/arm/lib/findbit.S | 78 ++++++++++++++++++-------------- 2 files changed, 50 insertions(+), 34 deletions(-) diff --git a/arch/arm/include/asm/assembler.h b/arch/arm/include/asm/assemb= ler.h index 90fbe4a3f9c8..28e18f79c300 100644 --- a/arch/arm/include/asm/assembler.h +++ b/arch/arm/include/asm/assembler.h @@ -761,6 +761,12 @@ THUMB( orr \reg , \reg , #PSR_T_BIT ) .endif .endm =20 + .if __LINUX_ARM_ARCH__ < 6 + .set .Lrev_l_uses_tmp, 1 + .else + .set .Lrev_l_uses_tmp, 0 + .endif + /* * bl_r - branch and link to register * diff --git a/arch/arm/lib/findbit.S b/arch/arm/lib/findbit.S index 8280f66d38a5..6ec584d16d46 100644 --- a/arch/arm/lib/findbit.S +++ b/arch/arm/lib/findbit.S @@ -14,32 +14,32 @@ #include .text =20 +#ifdef __ARMEB__ +#define SWAB_ENDIAN le +#else +#define SWAB_ENDIAN be +#endif + .macro find_first, endian, set, name ENTRY(_find_first_\name\()bit_\endian) teq r1, #0 beq 3f mov r2, #0 -1: - .ifc \endian, be - eor r3, r2, #0x18 - ARM( ldrb r3, [r0, r3, lsr #3] ) - THUMB( lsr r3, #3 ) - THUMB( ldrb r3, [r0, r3] ) +1: ldr r3, [r0], #4 + .ifeq \set + mvns r3, r3 @ invert/test bits .else - ARM( ldrb r3, [r0, r2, lsr #3] ) - THUMB( lsr r3, r2, #3 ) - THUMB( ldrb r3, [r0, r3] ) + movs r3, r3 @ test bits .endif - .ifeq \set - eors r3, r3, #0xff @ invert bits + .ifc \endian, SWAB_ENDIAN + bne .L_found_swab .else - movs r3, r3 + bne .L_found @ found the bit? .endif - bne .L_found @ any now set - found zero bit - add r2, r2, #8 @ next bit pointer + add r2, r2, #32 @ next index 2: cmp r2, r1 @ any more? blo 1b -3: mov r0, r1 @ no free bits +3: mov r0, r1 @ no more bits ret lr ENDPROC(_find_first_\name\()bit_\endian) .endm @@ -48,24 +48,25 @@ ENDPROC(_find_first_\name\()bit_\endian) ENTRY(_find_next_\name\()bit_\endian) cmp r2, r1 bhs 3b - ands ip, r2, #7 - beq 1b @ If new byte, goto old routine - .ifc \endian, be - eor r3, r2, #0x18 - ARM( ldrb r3, [r0, r3, lsr #3] ) - THUMB( lsr r3, #3 ) - THUMB( ldrb r3, [r0, r3] ) - .else - ARM( ldrb r3, [r0, r2, lsr #3] ) - THUMB( lsr r3, r2, #3 ) - THUMB( ldrb r3, [r0, r3] ) + mov ip, r2, lsr #5 @ word index + add r0, r0, ip, lsl #2 + ands ip, r2, #31 @ bit position + beq 1b + ldr r3, [r0], #4 + .ifeq \set + mvn r3, r3 @ invert bits + .endif + .ifc \endian, SWAB_ENDIAN + rev_l r3, ip + .if .Lrev_l_uses_tmp + @ we need to recompute ip because rev_l will have overwritten + @ it. + and ip, r2, #31 @ bit position .endif - .ifeq \set - eor r3, r3, #0xff @ now looking for a 1 bit .endif movs r3, r3, lsr ip @ shift off unused bits bne .L_found - orr r2, r2, #7 @ if zero, then no bits here + orr r2, r2, #31 @ no zero bits add r2, r2, #1 @ align bit pointer b 2b @ loop for next bit ENDPROC(_find_next_\name\()bit_\endian) @@ -95,6 +96,8 @@ ENDPROC(_find_next_\name\()bit_\endian) /* * One or more bits in the LSB of r3 are assumed to be set. */ +.L_found_swab: + rev_l r3, ip .L_found: #if __LINUX_ARM_ARCH__ >=3D 7 rbit r3, r3 @ reverse bits @@ -107,13 +110,20 @@ ENDPROC(_find_next_\name\()bit_\endian) rsb r3, r3, #31 @ offset of first set bit add r0, r2, r3 @ add offset of first set bit #else - tst r3, #0x0f + mov ip, #~0 + tst r3, ip, lsr #16 @ test bits 0-15 + addeq r2, r2, #16 + moveq r3, r3, lsr #16 + tst r3, #0x00ff + addeq r2, r2, #8 + moveq r3, r3, lsr #8 + tst r3, #0x000f addeq r2, r2, #4 - movne r3, r3, lsl #4 - tst r3, #0x30 + moveq r3, r3, lsr #4 + tst r3, #0x0003 addeq r2, r2, #2 - movne r3, r3, lsl #2 - tst r3, #0x40 + moveq r3, r3, lsr #2 + tst r3, #0x0001 addeq r2, r2, #1 mov r0, r2 #endif --=20 2.30.2 From nobody Sat Feb 7 15:16:54 2026 Return-Path: X-Spam-Checker-Version: SpamAssassin 3.4.0 (2014-02-07) on aws-us-west-2-korg-lkml-1.web.codeaurora.org Received: from vger.kernel.org (vger.kernel.org [23.128.96.18]) by smtp.lore.kernel.org (Postfix) with ESMTP id 9E91FECAAA1 for ; Fri, 28 Oct 2022 16:48:51 +0000 (UTC) Received: (majordomo@vger.kernel.org) by vger.kernel.org via listexpand id S230224AbiJ1Qss (ORCPT ); Fri, 28 Oct 2022 12:48:48 -0400 Received: from lindbergh.monkeyblade.net ([23.128.96.19]:39440 "EHLO lindbergh.monkeyblade.net" rhost-flags-OK-OK-OK-OK) by vger.kernel.org with ESMTP id S230085AbiJ1Qsa (ORCPT ); Fri, 28 Oct 2022 12:48:30 -0400 Received: from pandora.armlinux.org.uk (pandora.armlinux.org.uk [IPv6:2001:4d48:ad52:32c8:5054:ff:fe00:142]) by lindbergh.monkeyblade.net (Postfix) with ESMTPS id 33AC521464C for ; Fri, 28 Oct 2022 09:48:16 -0700 (PDT) DKIM-Signature: v=1; a=rsa-sha256; q=dns/txt; c=relaxed/relaxed; d=armlinux.org.uk; s=pandora-2019; h=Date:Sender:Message-Id:Content-Type: Content-Transfer-Encoding:MIME-Version:Subject:Cc:To:From:References: In-Reply-To:Reply-To:Content-ID:Content-Description:Resent-Date:Resent-From: Resent-Sender:Resent-To:Resent-Cc:Resent-Message-ID:List-Id:List-Help: List-Unsubscribe:List-Subscribe:List-Post:List-Owner:List-Archive; bh=OhpOuWjeen7QG1He6lXTiBfyZaGo2Ecqc0tyM0FhmYY=; b=v89NSJvDt4HVfbfHi/VcP+62QL Ofplk8KUS2qWPd42RjzNexz3DmjfBpg8PzD9c2v5GuTPyzMoFy44N3/Jv78HT3WE/uiilcEzYHUxQ o1oc4SaoCnGcCwSKBoeIw6NMgj5gLP3EJ5lWbz+FDcU8raSg+IWRv/Bziz+7+wfnXvw4TVpS0MdY1 dtIXLR7Od74TtK6+5kQwWuxtbmXh6HJQPaSWikgd4+0OgyQt0R6KNE0swC+qYMfsPZstYOk6ais83 WwVSMXbUj2Pn3UJaAtcO0IJKnUV+0HGqa9qnyzPNaC56T8Hyd/CkW2lWmXgE8Qn/KGoPmb7VL5vq7 q/JkKMIg==; Received: from e0022681537dd.dyn.armlinux.org.uk ([fd8f:7570:feb6:1:222:68ff:fe15:37dd]:40014 helo=rmk-PC.armlinux.org.uk) by pandora.armlinux.org.uk with esmtpsa (TLS1.3) tls TLS_ECDHE_RSA_WITH_AES_256_GCM_SHA384 (Exim 4.94.2) (envelope-from ) id 1ooSWk-00007E-6D; Fri, 28 Oct 2022 17:48:14 +0100 Received: from rmk by rmk-PC.armlinux.org.uk with local (Exim 4.94.2) (envelope-from ) id 1ooSWj-000FEM-Jl; Fri, 28 Oct 2022 17:48:13 +0100 In-Reply-To: References: From: "Russell King (Oracle)" To: Yury Norov Cc: Catalin Marinas , Mark Rutland , Will Deacon , Linux Kernel Mailing List , Linus Torvalds , linux-arm-kernel@lists.infradead.org Subject: [PATCH 5/5] ARM: findbit: add unwinder information MIME-Version: 1.0 Content-Disposition: inline Content-Transfer-Encoding: quoted-printable Content-Type: text/plain; charset="utf-8" Message-Id: Sender: Russell King Date: Fri, 28 Oct 2022 17:48:13 +0100 Precedence: bulk List-ID: X-Mailing-List: linux-kernel@vger.kernel.org Add unwinder information so oops in the findbit functions can create a proper backtrace. Signed-off-by: Russell King (Oracle) --- arch/arm/lib/findbit.S | 8 +++++++- 1 file changed, 7 insertions(+), 1 deletion(-) diff --git a/arch/arm/lib/findbit.S b/arch/arm/lib/findbit.S index 6ec584d16d46..b7ac2d3c0748 100644 --- a/arch/arm/lib/findbit.S +++ b/arch/arm/lib/findbit.S @@ -12,6 +12,7 @@ */ #include #include +#include .text =20 #ifdef __ARMEB__ @@ -22,6 +23,7 @@ =20 .macro find_first, endian, set, name ENTRY(_find_first_\name\()bit_\endian) + UNWIND( .fnstart) teq r1, #0 beq 3f mov r2, #0 @@ -41,11 +43,13 @@ ENTRY(_find_first_\name\()bit_\endian) blo 1b 3: mov r0, r1 @ no more bits ret lr + UNWIND( .fnend) ENDPROC(_find_first_\name\()bit_\endian) .endm =20 .macro find_next, endian, set, name ENTRY(_find_next_\name\()bit_\endian) + UNWIND( .fnstart) cmp r2, r1 bhs 3b mov ip, r2, lsr #5 @ word index @@ -69,6 +73,7 @@ ENTRY(_find_next_\name\()bit_\endian) orr r2, r2, #31 @ no zero bits add r2, r2, #1 @ align bit pointer b 2b @ loop for next bit + UNWIND( .fnend) ENDPROC(_find_next_\name\()bit_\endian) .endm =20 @@ -97,6 +102,7 @@ ENDPROC(_find_next_\name\()bit_\endian) * One or more bits in the LSB of r3 are assumed to be set. */ .L_found_swab: + UNWIND( .fnstart) rev_l r3, ip .L_found: #if __LINUX_ARM_ARCH__ >=3D 7 @@ -130,4 +136,4 @@ ENDPROC(_find_next_\name\()bit_\endian) cmp r1, r0 @ Clamp to maxbit movlo r0, r1 ret lr - + UNWIND( .fnend) --=20 2.30.2