From patchwork Wed May 29 14:53:46 2019 Content-Type: text/plain; charset="utf-8" MIME-Version: 1.0 Content-Transfer-Encoding: 8bit X-Patchwork-Submitter: Florian Weimer X-Patchwork-Id: 1107202 Return-Path: X-Original-To: incoming@patchwork.ozlabs.org Delivered-To: patchwork-incoming@bilbo.ozlabs.org Authentication-Results: ozlabs.org; spf=pass (mailfrom) smtp.mailfrom=sourceware.org (client-ip=209.132.180.131; helo=sourceware.org; envelope-from=libc-alpha-return-102339-incoming=patchwork.ozlabs.org@sourceware.org; receiver=) Authentication-Results: ozlabs.org; dmarc=fail (p=none dis=none) header.from=redhat.com Authentication-Results: ozlabs.org; dkim=pass (1024-bit key; secure) header.d=sourceware.org header.i=@sourceware.org header.b="dq7DoZ0x"; dkim-atps=neutral Received: from sourceware.org (server1.sourceware.org [209.132.180.131]) (using TLSv1.2 with cipher ECDHE-RSA-AES256-GCM-SHA384 (256/256 bits)) (No client certificate requested) by ozlabs.org (Postfix) with ESMTPS id 45DYd25hDNz9s3l for ; Thu, 30 May 2019 00:53:58 +1000 (AEST) DomainKey-Signature: a=rsa-sha1; c=nofws; d=sourceware.org; h=list-id :list-unsubscribe:list-subscribe:list-archive:list-post :list-help:sender:from:to:cc:subject:references:date:in-reply-to :message-id:mime-version:content-type:content-transfer-encoding; q=dns; s=default; b=RA9CDp8H1Nl6iAqPdd/uB8NZusGy/pZjoOFDm+539Sn AlYojVR8zbZzxPzmMpeNmw+nTTJFWq7zrzPqRI1v0JHJQG9JOQhAzxaNJQQgwzfj Q2orULXoyF3E/H8oAqpux3zkG3nExrB5p1ez99d1jmmQJycBer7v7fbDXqCmkmrI = DKIM-Signature: v=1; a=rsa-sha1; c=relaxed; d=sourceware.org; h=list-id :list-unsubscribe:list-subscribe:list-archive:list-post :list-help:sender:from:to:cc:subject:references:date:in-reply-to :message-id:mime-version:content-type:content-transfer-encoding; s=default; bh=RjfTeUeaaoUZtLBQTY5f7fC7KiQ=; b=dq7DoZ0xdczPtl0UI UbXEhrS5/nQfux0N86K74MBIjeg/4xC5rtv3cg7aWkiE187TgOq7bw3G7m5U3PoY M5q2cH8iLFgxKDIftdSdnJrv7guAcVLgqMbVa9JQtfH6OHC8RAvCiW59DJVKTMEb gQzQi6DwqhjgS5irKu5zWZWliE= Received: (qmail 57503 invoked by alias); 29 May 2019 14:53:52 -0000 Mailing-List: contact libc-alpha-help@sourceware.org; run by ezmlm Precedence: bulk List-Id: List-Unsubscribe: List-Subscribe: List-Archive: List-Post: List-Help: , Sender: libc-alpha-owner@sourceware.org Delivered-To: mailing list libc-alpha@sourceware.org Received: (qmail 57494 invoked by uid 89); 29 May 2019 14:53:52 -0000 Authentication-Results: sourceware.org; auth=none X-Spam-SWARE-Status: No, score=-18.9 required=5.0 tests=AWL, BAYES_00, GIT_PATCH_0, GIT_PATCH_1, GIT_PATCH_2, GIT_PATCH_3, KAM_SHORT, SPF_HELO_PASS autolearn=ham version=3.3.1 spammy= X-HELO: mx1.redhat.com From: Florian Weimer To: Phil Blundell Cc: libc-alpha@sourceware.org Subject: [PATCH] Remove ioperm etc. support for arm (was: Re: _ioperm support for Arm) References: <87o93lo3y9.fsf@oldenburg2.str.redhat.com> <20190529135135.rt3c5fhhser7fik6@hetzner.pbcl.net> <875zptqrwx.fsf@oldenburg2.str.redhat.com> <20190529141820.5nny64ugrvqyzqgd@hetzner.pbcl.net> Date: Wed, 29 May 2019 16:53:46 +0200 In-Reply-To: <20190529141820.5nny64ugrvqyzqgd@hetzner.pbcl.net> (Phil Blundell's message of "Wed, 29 May 2019 16:18:20 +0200") Message-ID: <87r28hpb0l.fsf_-_@oldenburg2.str.redhat.com> User-Agent: Gnus/5.13 (Gnus v5.13) Emacs/26.2 (gnu/linux) MIME-Version: 1.0 * Phil Blundell: > On Wed, May 29, 2019 at 04:03:26PM +0200, Florian Weimer wrote: >> There's this bug: >> >> Remove ARM old-ABI support >> >> >> I believe this removed core support for non-EABI platforms. > > That's not quite the same thing. The old ABI differed from the EABI > in several ways, including the stack alignment rules and the way > that system calls were done. I think the bug you mentioned above > was to remove the support in glibc for those specific "old ABI" > mechanisms. > > What you currently get if you build glibc for ARMv4 is an "almost-EABI" > configuration which does everything in the same way as the EABI except > that there is no interworking support. I think we still have a certain > amount of #ifdef __ARM_ARCH_4__ for that purpose. If we wanted to clean > things up there then removing support for the half-baked V4T interworking > would probably be more of a win since I think that accounts for a larger > amount of #ifdef scar tissue. > >> All the target triplets we test upstream have gnueabi in them, and these >> days, anything that is not built by build-many-glibcs.py in at least >> variant is not considered supported, I think. > > For the reasons above, even when building glibc for armv4 you still put > "gnueabi" in the triplet. But I think you're still correct that > build-many-glibcs.py doesn't actually test anything older than ARMv5TE. For libc.so.6 on arm-linux-gnueabi (as built by build-many-glibcs.py), readelf shows this: Attribute Section: aeabi File Attributes Tag_CPU_name: "5T" Tag_CPU_arch: v5T Tag_ARM_ISA_use: Yes Tag_THUMB_ISA_use: Thumb-1 Tag_ABI_PCS_wchar_t: 4 Tag_ABI_FP_rounding: Needed Tag_ABI_FP_denormal: Needed Tag_ABI_FP_exceptions: Needed Tag_ABI_FP_user_exceptions: Needed Tag_ABI_FP_number_model: IEEE 754 Tag_ABI_align_needed: 8-byte Tag_ABI_align_preserved: 8-byte, except leaf SP Tag_ABI_enum_size: int So it looks like to me like a later architectue version. I would welcome comments on the commit message. The patch itself has been “tested” with build-many-glibcs.py on all 32-bit Arm architectures. Thanks, Florian arm: Remove ioperm/iopl/inb/inw/inl/outb/outw/outl support Linux only supports the required ISA sysctls on StrongARM devices, which are armv4 and no longer tested during glibc development and probably bit-rotted by this point. (No reported test results, and the last discussion of armv4 support was in the glibc 2.19 release notes.) 2019-05-29 Florian Weimer arm: Remove ioperm/iopl/inb/inw/inl/outb/outw/outl support. * sysdeps/unix/sysv/linux/arm/Makefile [$(subdir) == misc] (sysdep_headers): Remove sys/io.h. * sysdeps/unix/sysv/linux/arm/sys/io.h: Remove file. * sysdeps/unix/sysv/linux/arm/ioperm.c: Rewrite file. (ioperm, iopl, inb, inw, inl, outb, outw, outl): Turn into compatibility symbols. diff --git a/NEWS b/NEWS index c885b960ca..31ead24851 100644 --- a/NEWS +++ b/NEWS @@ -58,6 +58,9 @@ Deprecated and removed features, and other changes affecting compatibility: configurations) has been removed, following the deprecation of this subarchitecture in version 8 of GCC, and its removal in version 9. +* On 32-bit Arm, support for the port-based I/O emulation and the + header have been removed. + Changes to build and runtime requirements: * GCC 6.2 or later is required to build the GNU C Library. diff --git a/sysdeps/unix/sysv/linux/arm/Makefile b/sysdeps/unix/sysv/linux/arm/Makefile index 4adc35de04..d7a2f6a8a7 100644 --- a/sysdeps/unix/sysv/linux/arm/Makefile +++ b/sysdeps/unix/sysv/linux/arm/Makefile @@ -5,7 +5,7 @@ endif ifeq ($(subdir),misc) sysdep_routines += ioperm -sysdep_headers += sys/elf.h sys/io.h +sysdep_headers += sys/elf.h endif ifeq ($(subdir),signal) diff --git a/sysdeps/unix/sysv/linux/arm/ioperm.c b/sysdeps/unix/sysv/linux/arm/ioperm.c index 0e338c4504..e832a6605e 100644 --- a/sysdeps/unix/sysv/linux/arm/ioperm.c +++ b/sysdeps/unix/sysv/linux/arm/ioperm.c @@ -17,167 +17,71 @@ License along with the GNU C Library. If not, see . */ -/* I/O port access on the ARM is something of a fiction. What we do is to - map an appropriate area of /dev/mem into user space so that a program - can blast away at the hardware in such a way as to generate I/O cycles - on the bus. To insulate user code from dependencies on particular - hardware we don't allow calls to inb() and friends to be inlined, but - force them to come through code in here every time. Performance-critical - registers tend to be memory mapped these days so this should be no big - problem. */ +#include -/* Once upon a time this file used mprotect to enable and disable - access to particular areas of I/O space. Unfortunately the - mprotect syscall also has the side effect of enabling caching for - the area affected (this is a kernel limitation). So we now just - enable all the ports all of the time. */ +#if SHLIB_COMPAT (libc, GLIBC_2_4, GLIBC_2_30) +# include -#include -#include -#include -#include -#include -#include - -#include -#include - -#include - -#define MAX_PORT 0x10000 - -static struct { - unsigned long int base; - unsigned long int io_base; - unsigned int shift; - unsigned int initdone; /* since all the above could be 0 */ -} io; - -#define IO_ADDR(port) (io.base + ((port) << io.shift)) - -/* - * Initialize I/O system. The io_bae and port_shift values are fetched - * using sysctl (CTL_BUS, CTL_BUS_ISA, ISA_*). - */ - -static int -init_iosys (void) +int +ioperm (unsigned long int from, unsigned long int num, int turn_on) { - static int iobase_name[] = { CTL_BUS, CTL_BUS_ISA, BUS_ISA_PORT_BASE }; - static int ioshift_name[] = { CTL_BUS, CTL_BUS_ISA, BUS_ISA_PORT_SHIFT }; - size_t len = sizeof (io.base); - - if (! __sysctl (iobase_name, 3, &io.io_base, &len, NULL, 0) - && ! __sysctl (ioshift_name, 3, &io.shift, &len, NULL, 0)) - { - io.initdone = 1; - return 0; - } - - /* sysctl has failed... */ - __set_errno (ENODEV); + __set_errno (ENOSYS); return -1; } +compat_symbol (libc, ioperm, ioperm, GLIBC_2_4); int -_ioperm (unsigned long int from, unsigned long int num, int turn_on) +iopl (unsigned int level) { - if (! io.initdone && init_iosys () < 0) - return -1; - - /* this test isn't as silly as it may look like; consider overflows! */ - if (from >= MAX_PORT || from + num > MAX_PORT) - { - __set_errno (EINVAL); - return -1; - } - - if (turn_on) - { - if (! io.base) - { - int fd; - - fd = __open ("/dev/mem", O_RDWR); - if (fd < 0) - return -1; - - io.base = - (unsigned long int) __mmap (0, MAX_PORT << io.shift, - PROT_READ | PROT_WRITE, - MAP_SHARED, fd, io.io_base); - __close (fd); - if ((long) io.base == -1) - return -1; - } - } - - return 0; + __set_errno (ENOSYS); + return -1; } +compat_symbol (libc, iopl, iopl, GLIBC_2_4); -int -_iopl (unsigned int level) -{ - if (level > 3) - { - __set_errno (EINVAL); - return -1; - } - if (level) - { - return _ioperm (0, MAX_PORT, 1); - } - return 0; -} - +/* The remaining functions do not have any way to indicate failure. + However, it is only valid to call them after calling ioperm/iopl, + which will have indicated failure. */ void -_outb (unsigned char b, unsigned long int port) +outb (unsigned char b, unsigned long int port) { - *((volatile unsigned char *)(IO_ADDR (port))) = b; } - +compat_symbol (libc, outb, outb, GLIBC_2_4); void -_outw (unsigned short b, unsigned long int port) +outw (unsigned short b, unsigned long int port) { - *((volatile unsigned short *)(IO_ADDR (port))) = b; } - +compat_symbol (libc, outw, outw, GLIBC_2_4); void -_outl (unsigned int b, unsigned long int port) +outl (unsigned int b, unsigned long int port) { - *((volatile unsigned long *)(IO_ADDR (port))) = b; } - +compat_symbol (libc, outl, outl, GLIBC_2_4); unsigned int -_inb (unsigned long int port) +inb (unsigned long int port) { - return *((volatile unsigned char *)(IO_ADDR (port))); + return 0; } +compat_symbol (libc, inb, inb, GLIBC_2_4); unsigned int -_inw (unsigned long int port) +inw (unsigned long int port) { - return *((volatile unsigned short *)(IO_ADDR (port))); + return 0; } +compat_symbol (libc, inw, inw, GLIBC_2_4); unsigned int -_inl (unsigned long int port) +inl (unsigned long int port) { - return *((volatile unsigned long *)(IO_ADDR (port))); + return 0; } +compat_symbol (libc, inl, inl, GLIBC_2_4); -weak_alias (_ioperm, ioperm); -weak_alias (_iopl, iopl); -weak_alias (_inb, inb); -weak_alias (_inw, inw); -weak_alias (_inl, inl); -weak_alias (_outb, outb); -weak_alias (_outw, outw); -weak_alias (_outl, outl); +#endif /* SHLIB_COMAT */ diff --git a/sysdeps/unix/sysv/linux/arm/sys/io.h b/sysdeps/unix/sysv/linux/arm/sys/io.h deleted file mode 100644 index 6f5ae5079a..0000000000 --- a/sysdeps/unix/sysv/linux/arm/sys/io.h +++ /dev/null @@ -1,47 +0,0 @@ -/* Copyright (C) 1996-2019 Free Software Foundation, Inc. - This file is part of the GNU C Library. - - The GNU C Library is free software; you can redistribute it and/or - modify it under the terms of the GNU Lesser General Public - License as published by the Free Software Foundation; either - version 2.1 of the License, or (at your option) any later version. - - The GNU C Library is distributed in the hope that it will be useful, - but WITHOUT ANY WARRANTY; without even the implied warranty of - MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU - Lesser General Public License for more details. - - You should have received a copy of the GNU Lesser General Public - License along with the GNU C Library. If not, see - . */ - -#ifndef _SYS_IO_H - -#define _SYS_IO_H 1 -#include - -__BEGIN_DECLS - -/* If TURN_ON is TRUE, request for permission to do direct i/o on the - port numbers in the range [FROM,FROM+NUM-1]. Otherwise, turn I/O - permission off for that range. This call requires root privileges. */ -extern int ioperm (unsigned long int __from, unsigned long int __num, - int __turn_on) __THROW; - -/* Set the I/O privilege level to LEVEL. If LEVEL is nonzero, - permission to access any I/O port is granted. This call requires - root privileges. */ -extern int iopl (int __level) __THROW; - -/* The functions that actually perform reads and writes. */ -extern unsigned char inb (unsigned long int __port) __THROW; -extern unsigned short int inw (unsigned long int __port) __THROW; -extern unsigned long int inl (unsigned long int __port) __THROW; - -extern void outb (unsigned char __value, unsigned long int __port) __THROW; -extern void outw (unsigned short __value, unsigned long int __port) __THROW; -extern void outl (unsigned long __value, unsigned long int __port) __THROW; - -__END_DECLS - -#endif /* _SYS_IO_H */