From mboxrd@z Thu Jan 1 00:00:00 1970 Return-Path: Received: from hall.aurel32.net (hall.aurel32.net [IPv6:2001:bc8:30d7:100::1]) by sourceware.org (Postfix) with ESMTPS id 44B00393FC2F for ; Wed, 11 Mar 2020 12:03:01 +0000 (GMT) Received: from [2a01:e35:2fdd:a4e1:fe91:fc89:bc43:b814] (helo=ohm.rr44.fr) by hall.aurel32.net with esmtpsa (TLS1.3:ECDHE_RSA_AES_256_GCM_SHA384:256) (Exim 4.92) (envelope-from ) id 1jC04h-0004Pn-P5; Wed, 11 Mar 2020 13:02:59 +0100 Received: from aurel32 by ohm.rr44.fr with local (Exim 4.93) (envelope-from ) id 1jC04h-004p88-4F; Wed, 11 Mar 2020 13:02:59 +0100 From: Aurelien Jarno To: libc-stable@sourceware.org Cc: Joseph Myers Subject: [2.30 COMMITTED 1/3] Avoid ldbl-96 stack corruption from range reduction of pseudo-zero (bug 25487). Date: Wed, 11 Mar 2020 13:02:43 +0100 Message-Id: <20200311120245.1026089-1-aurelien@aurel32.net> X-Mailer: git-send-email 2.24.1 MIME-Version: 1.0 Content-Transfer-Encoding: 8bit X-Spam-Status: No, score=-27.6 required=5.0 tests=BAYES_00, GIT_PATCH_0, GIT_PATCH_1, GIT_PATCH_2, GIT_PATCH_3, RCVD_IN_DNSWL_LOW, SPF_HELO_PASS, SPF_NONE autolearn=ham autolearn_force=no version=3.4.2 X-Spam-Checker-Version: SpamAssassin 3.4.2 (2018-09-13) on server2.sourceware.org X-BeenThere: libc-stable@sourceware.org X-Mailman-Version: 2.1.29 Precedence: list List-Id: Libc-stable mailing list List-Unsubscribe: , List-Archive: List-Help: List-Subscribe: , X-List-Received-Date: Wed, 11 Mar 2020 12:03:02 -0000 From: Joseph Myers Bug 25487 reports stack corruption in ldbl-96 sinl on a pseudo-zero argument (an representation where all the significand bits, including the explicit high bit, are zero, but the exponent is not zero, which is not a valid representation for the long double type). Although this is not a valid long double representation, existing practice in this area (see bug 4586, originally marked invalid but subsequently fixed) is that we still seek to avoid invalid memory accesses as a result, in case of programs that treat arbitrary binary data as long double representations, although the invalid representations of the ldbl-96 format do not need to be consistently handled the same as any particular valid representation. This patch makes the range reduction detect pseudo-zero and unnormal representations that would otherwise go to __kernel_rem_pio2, and returns a NaN for them instead of continuing with the range reduction process. (Pseudo-zero and unnormal representations whose unbiased exponent is less than -1 have already been safely returned from the function before this point without going through the rest of range reduction.) Pseudo-zero representations would previously result in the value passed to __kernel_rem_pio2 being all-zero, which is definitely unsafe; unnormal representations would previously result in a value passed whose high bit is zero, which might well be unsafe since that is not a form of input expected by __kernel_rem_pio2. Tested for x86_64. (cherry picked from commit 9333498794cde1d5cca518badf79533a24114b6f) --- NEWS | 1 + sysdeps/ieee754/ldbl-96/Makefile | 3 +- sysdeps/ieee754/ldbl-96/e_rem_pio2l.c | 12 +++++++ sysdeps/ieee754/ldbl-96/test-sinl-pseudo.c | 41 ++++++++++++++++++++++ 4 files changed, 56 insertions(+), 1 deletion(-) create mode 100644 sysdeps/ieee754/ldbl-96/test-sinl-pseudo.c diff --git a/NEWS b/NEWS index 555d3b9a093..67018b06582 100644 --- a/NEWS +++ b/NEWS @@ -35,6 +35,7 @@ The following bugs are resolved with this release: [25225] ld.so fails to link on x86 if GCC defaults to -fcf-protection [25232] No const correctness for strchr et al. for Clang++ [25401] Remove incorrect alloc_size attribute from pvalloc + [25487] sinl() stack corruption from crafted input (CVE-2020-10029) [25523] MIPS/Linux inline syscall template is miscompiled diff --git a/sysdeps/ieee754/ldbl-96/Makefile b/sysdeps/ieee754/ldbl-96/Makefile index b1032542140..052c1c77037 100644 --- a/sysdeps/ieee754/ldbl-96/Makefile +++ b/sysdeps/ieee754/ldbl-96/Makefile @@ -17,5 +17,6 @@ # . ifeq ($(subdir),math) -tests += test-canonical-ldbl-96 test-totalorderl-ldbl-96 +tests += test-canonical-ldbl-96 test-totalorderl-ldbl-96 test-sinl-pseudo +CFLAGS-test-sinl-pseudo.c += -fstack-protector-all endif diff --git a/sysdeps/ieee754/ldbl-96/e_rem_pio2l.c b/sysdeps/ieee754/ldbl-96/e_rem_pio2l.c index 805de22d732..1aeccb47d72 100644 --- a/sysdeps/ieee754/ldbl-96/e_rem_pio2l.c +++ b/sysdeps/ieee754/ldbl-96/e_rem_pio2l.c @@ -210,6 +210,18 @@ __ieee754_rem_pio2l (long double x, long double *y) return 0; } + if ((i0 & 0x80000000) == 0) + { + /* Pseudo-zero and unnormal representations are not valid + representations of long double. We need to avoid stack + corruption in __kernel_rem_pio2, which expects input in a + particular normal form, but those representations do not need + to be consistently handled like any particular floating-point + value. */ + y[1] = y[0] = __builtin_nanl (""); + return 0; + } + /* Split the 64 bits of the mantissa into three 24-bit integers stored in a double array. */ exp = j0 - 23; diff --git a/sysdeps/ieee754/ldbl-96/test-sinl-pseudo.c b/sysdeps/ieee754/ldbl-96/test-sinl-pseudo.c new file mode 100644 index 00000000000..f59b97769d0 --- /dev/null +++ b/sysdeps/ieee754/ldbl-96/test-sinl-pseudo.c @@ -0,0 +1,41 @@ +/* Test sinl for pseudo-zeros and unnormals for ldbl-96 (bug 25487). + Copyright (C) 2020 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 + . */ + +#include +#include +#include + +static int +do_test (void) +{ + for (int i = 0; i < 64; i++) + { + uint64_t sig = i == 63 ? 0 : 1ULL << i; + long double ld; + SET_LDOUBLE_WORDS (ld, 0x4141, + sig >> 32, sig & 0xffffffffULL); + /* The requirement is that no stack overflow occurs when the + pseudo-zero or unnormal goes through range reduction. */ + volatile long double ldr; + ldr = sinl (ld); + (void) ldr; + } + return 0; +} + +#include -- 2.24.1