From mboxrd@z Thu Jan 1 00:00:00 1970 Return-Path: Received: from us-smtp-delivery-124.mimecast.com (us-smtp-delivery-124.mimecast.com [170.10.129.124]) by sourceware.org (Postfix) with ESMTPS id 94340385842B for ; Wed, 27 Apr 2022 18:00:40 +0000 (GMT) DMARC-Filter: OpenDMARC Filter v1.4.1 sourceware.org 94340385842B Received: from mimecast-mx02.redhat.com (mimecast-mx02.redhat.com [66.187.233.88]) by relay.mimecast.com with ESMTP with STARTTLS (version=TLSv1.2, cipher=TLS_ECDHE_RSA_WITH_AES_256_GCM_SHA384) id us-mta-461-5UiNxADvOAqcYsTWqdoeeA-1; Wed, 27 Apr 2022 14:00:39 -0400 X-MC-Unique: 5UiNxADvOAqcYsTWqdoeeA-1 Received: from smtp.corp.redhat.com (int-mx08.intmail.prod.int.rdu2.redhat.com [10.11.54.8]) (using TLSv1.2 with cipher AECDH-AES256-SHA (256/256 bits)) (No client certificate requested) by mimecast-mx02.redhat.com (Postfix) with ESMTPS id AF2AC1014A60; Wed, 27 Apr 2022 18:00:38 +0000 (UTC) Received: from oldenburg.str.redhat.com (unknown [10.39.193.187]) by smtp.corp.redhat.com (Postfix) with ESMTPS id B7934C28118; Wed, 27 Apr 2022 18:00:32 +0000 (UTC) From: Florian Weimer To: Adhemerval Zanella Cc: Adhemerval Zanella via Libc-alpha Subject: Re: [PATCH v2 2/4] Consolidate stdio-lock.h References: <20220426191523.833171-1-adhemerval.zanella@linaro.org> <20220426191523.833171-3-adhemerval.zanella@linaro.org> <87o80mhd4u.fsf@oldenburg.str.redhat.com> Date: Wed, 27 Apr 2022 20:00:30 +0200 In-Reply-To: (Adhemerval Zanella's message of "Wed, 27 Apr 2022 13:15:18 -0300") Message-ID: <874k2ee79d.fsf@oldenburg.str.redhat.com> User-Agent: Gnus/5.13 (Gnus v5.13) Emacs/27.2 (gnu/linux) MIME-Version: 1.0 X-Scanned-By: MIMEDefang 2.85 on 10.11.54.8 X-Mimecast-Spam-Score: 0 X-Mimecast-Originator: redhat.com Content-Type: text/plain X-Spam-Status: No, score=-11.1 required=5.0 tests=BAYES_00, DKIMWL_WL_HIGH, DKIM_SIGNED, DKIM_VALID, DKIM_VALID_AU, DKIM_VALID_EF, GIT_PATCH_0, RCVD_IN_DNSWL_LOW, SPF_HELO_NONE, SPF_NONE, TXREP autolearn=ham autolearn_force=no version=3.4.4 X-Spam-Checker-Version: SpamAssassin 3.4.4 (2020-01-24) on server2.sourceware.org X-BeenThere: libc-alpha@sourceware.org X-Mailman-Version: 2.1.29 Precedence: list List-Id: Libc-alpha mailing list List-Unsubscribe: , List-Archive: List-Post: List-Help: List-Subscribe: , X-List-Received-Date: Wed, 27 Apr 2022 18:00:42 -0000 * Adhemerval Zanella: > On 27/04/2022 10:25, Florian Weimer wrote: >> * Adhemerval Zanella via Libc-alpha: >> >>> diff --git a/sysdeps/generic/stdio-lock.h b/sysdeps/generic/stdio-lock.h >>> index 14cf458bdd..fd61f0b5b7 100644 >>> --- a/sysdeps/generic/stdio-lock.h >>> +++ b/sysdeps/generic/stdio-lock.h >>> @@ -45,20 +45,13 @@ __libc_lock_define_recursive (typedef, _IO_lock_t) >>> #define _IO_cleanup_region_end(_doit) \ >>> __libc_cleanup_region_end (_doit) >>> >>> -#if defined _LIBC && IS_IN (libc) >>> - >>> -# ifdef __EXCEPTIONS >>> -# define _IO_acquire_lock(_fp) \ >>> +#define _IO_acquire_lock(_fp) \ >>> do { \ >>> - FILE *_IO_acquire_lock_file \ >>> - __attribute__((cleanup (_IO_acquire_lock_fct))) \ >>> - = (_fp); \ >>> - _IO_flockfile (_IO_acquire_lock_file); >>> -# else >>> -# define _IO_acquire_lock(_fp) _IO_acquire_lock_needs_exceptions_enabled >>> -# endif >>> -# define _IO_release_lock(_fp) ; } while (0) >>> - >>> -#endif >>> + _IO_cleanup_region_start((void (*) (void *)) &_IO_funlockfile, _fp); \ >>> + _IO_flockfile (_fp); >>> +#define _IO_release_lock(_fp) \ >>> + _IO_funlockfile (_fp); \ >>> + _IO_cleanup_region_end (0); \ >>> + } while (0) >>> >>> #endif /* stdio-lock.h */ >> >> I think this change replaces unwind tables for -fexceptions builds. If >> GCC can't turn the indirect call to the unlock function into a direct >> call, this will result in a loss of hardening due to the additional >> indirect function call. >> >> This change may also lose C++ unwinding compatibility for some >> fopencookie use cases, I think. > > This is an internal header where if __EXCEPTIONS is not defined we will > get a compiler error because of an undefined symbol > (_IO_acquire_lock_needs_exceptions_enabled). So internally all > _IO_acquire_lock usage already requires __EXCEPTIONS, so the fallback > is just unused definitions. I see this code generation change in libio/fputc.os. The new code uses an on-stack pointer saved at the start of the cleanup region: + 90: 48 8b 05 00 00 00 00 mov 0x0(%rip),%rax # 97 + 93: R_X86_64_REX_GOTPCRELX _IO_funlockfile-0x4 + 97: 48 89 e7 mov %rsp,%rdi + 9a: 48 89 04 24 mov %rax,(%rsp) + 9e: e8 00 00 00 00 call a3 + 9f: R_X86_64_PLT32 __GI___libc_cleanup_push_defer-0x4 + a3: 8b 45 00 mov 0x0(%rbp),%eax + a6: 25 00 80 00 00 and $0x8000,%eax + ab: 0f 85 d1 00 00 00 jne 182 + b1: 64 4c 8b 2c 25 10 00 mov %fs:0x10,%r13 + b8: 00 00 + ba: 48 8b bd 88 00 00 00 mov 0x88(%rbp),%rdi + c1: 4c 39 6f 08 cmp %r13,0x8(%rdi) + c5: 74 1a je e1 + c7: ba 01 00 00 00 mov $0x1,%edx + cc: f0 0f b1 17 lock cmpxchg %edx,(%rdi) + d0: 0f 85 a2 00 00 00 jne 178 + d6: 48 8b bd 88 00 00 00 mov 0x88(%rbp),%rdi + dd: 4c 89 6f 08 mov %r13,0x8(%rdi) + e1: 83 47 04 01 addl $0x1,0x4(%rdi) + e5: 41 bd 01 00 00 00 mov $0x1,%r13d + eb: e9 3d ff ff ff jmp 2d + f0: 48 89 e7 mov %rsp,%rdi + f3: e8 00 00 00 00 call f8 + f4: R_X86_64_PLT32 __GI___libc_cleanup_pop_restore-0x4 This is a from a build with CFLAGS="-O2 -fexceptions -s -DNDEBUG" (for comparison purposes). The old code just inlined the _IO_funlockfile fast path. (We seem to lack libc_hidden_proto/libc_hidden_def for _IO_funlockfile.) Thanks, Florian