Revision 472e5b056f000a778abb41f1e443de58eb259783 authored by Linus Torvalds on 02 October 2020, 02:14:36 UTC, committed by Linus Torvalds on 02 October 2020, 02:14:36 UTC
The pipe splice code still used the old model of waiting for pipe IO by
using a non-specific "pipe_wait()" that waited for any pipe event to
happen, which depended on all pipe IO being entirely serialized by the
pipe lock. So by checking the state you were waiting for, and then
adding yourself to the wait queue before dropping the lock, you were
guaranteed to see all the wakeups.
Strictly speaking, the actual wakeups were not done under the lock, but
the pipe_wait() model still worked, because since the waiter held the
lock when checking whether it should sleep, it would always see the
current state, and the wakeup was always done after updating the state.
However, commit 0ddad21d3e99 ("pipe: use exclusive waits when reading or
writing") split the single wait-queue into two, and in the process also
made the "wait for event" code wait for _two_ wait queues, and that then
showed a race with the wakers that were not serialized by the pipe lock.
It's only splice that used that "pipe_wait()" model, so the problem
wasn't obvious, but Josef Bacik reports:
"I hit a hang with fstest btrfs/187, which does a btrfs send into
/dev/null. This works by creating a pipe, the write side is given to
the kernel to write into, and the read side is handed to a thread that
splices into a file, in this case /dev/null.
The box that was hung had the write side stuck here [pipe_write] and
the read side stuck here [splice_from_pipe_next -> pipe_wait].
[ more details about pipe_wait() scenario ]
The problem is we're doing the prepare_to_wait, which sets our state
each time, however we can be woken up either with reads or writes. In
the case above we race with the WRITER waking us up, and re-set our
state to INTERRUPTIBLE, and thus never break out of schedule"
Josef had a patch that avoided the issue in pipe_wait() by just making
it set the state only once, but the deeper problem is that pipe_wait()
depends on a level of synchonization by the pipe mutex that it really
shouldn't. And the whole "wait for any pipe state change" model really
isn't very good to begin with.
So rather than trying to work around things in pipe_wait(), remove that
legacy model of "wait for arbitrary pipe event" entirely, and actually
create functions that wait for the pipe actually being readable or
writable, and can do so without depending on the pipe lock serializing
everything.
Fixes: 0ddad21d3e99 ("pipe: use exclusive waits when reading or writing")
Link: https://lore.kernel.org/linux-fsdevel/bfa88b5ad6f069b2b679316b9e495a970130416c.1601567868.git.josef@toxicpanda.com/
Reported-by: Josef Bacik <josef@toxicpanda.com>
Reviewed-and-tested-by: Josef Bacik <josef@toxicpanda.com>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
1 parent 44b6e23
ecrdsa_defs.h
/* SPDX-License-Identifier: GPL-2.0+ */
/*
* Definitions of EC-RDSA Curve Parameters
*
* Copyright (c) 2019 Vitaly Chikunov <vt@altlinux.org>
*
* This program is free software; you can redistribute it and/or modify it
* under the terms of the GNU General Public License as published by the Free
* Software Foundation; either version 2 of the License, or (at your option)
* any later version.
*/
#ifndef _CRYTO_ECRDSA_DEFS_H
#define _CRYTO_ECRDSA_DEFS_H
#include "ecc.h"
#define ECRDSA_MAX_SIG_SIZE (2 * 512 / 8)
#define ECRDSA_MAX_DIGITS (512 / 64)
/*
* EC-RDSA uses its own set of curves.
*
* cp256{a,b,c} curves first defined for GOST R 34.10-2001 in RFC 4357 (as
* 256-bit {A,B,C}-ParamSet), but inherited for GOST R 34.10-2012 and
* proposed for use in R 50.1.114-2016 and RFC 7836 as the 256-bit curves.
*/
/* OID_gostCPSignA 1.2.643.2.2.35.1 */
static u64 cp256a_g_x[] = {
0x0000000000000001ull, 0x0000000000000000ull,
0x0000000000000000ull, 0x0000000000000000ull, };
static u64 cp256a_g_y[] = {
0x22ACC99C9E9F1E14ull, 0x35294F2DDF23E3B1ull,
0x27DF505A453F2B76ull, 0x8D91E471E0989CDAull, };
static u64 cp256a_p[] = { /* p = 2^256 - 617 */
0xFFFFFFFFFFFFFD97ull, 0xFFFFFFFFFFFFFFFFull,
0xFFFFFFFFFFFFFFFFull, 0xFFFFFFFFFFFFFFFFull };
static u64 cp256a_n[] = {
0x45841B09B761B893ull, 0x6C611070995AD100ull,
0xFFFFFFFFFFFFFFFFull, 0xFFFFFFFFFFFFFFFFull };
static u64 cp256a_a[] = { /* a = p - 3 */
0xFFFFFFFFFFFFFD94ull, 0xFFFFFFFFFFFFFFFFull,
0xFFFFFFFFFFFFFFFFull, 0xFFFFFFFFFFFFFFFFull };
static u64 cp256a_b[] = {
0x00000000000000a6ull, 0x0000000000000000ull,
0x0000000000000000ull, 0x0000000000000000ull };
static struct ecc_curve gost_cp256a = {
.name = "cp256a",
.g = {
.x = cp256a_g_x,
.y = cp256a_g_y,
.ndigits = 256 / 64,
},
.p = cp256a_p,
.n = cp256a_n,
.a = cp256a_a,
.b = cp256a_b
};
/* OID_gostCPSignB 1.2.643.2.2.35.2 */
static u64 cp256b_g_x[] = {
0x0000000000000001ull, 0x0000000000000000ull,
0x0000000000000000ull, 0x0000000000000000ull, };
static u64 cp256b_g_y[] = {
0x744BF8D717717EFCull, 0xC545C9858D03ECFBull,
0xB83D1C3EB2C070E5ull, 0x3FA8124359F96680ull, };
static u64 cp256b_p[] = { /* p = 2^255 + 3225 */
0x0000000000000C99ull, 0x0000000000000000ull,
0x0000000000000000ull, 0x8000000000000000ull, };
static u64 cp256b_n[] = {
0xE497161BCC8A198Full, 0x5F700CFFF1A624E5ull,
0x0000000000000001ull, 0x8000000000000000ull, };
static u64 cp256b_a[] = { /* a = p - 3 */
0x0000000000000C96ull, 0x0000000000000000ull,
0x0000000000000000ull, 0x8000000000000000ull, };
static u64 cp256b_b[] = {
0x2F49D4CE7E1BBC8Bull, 0xE979259373FF2B18ull,
0x66A7D3C25C3DF80Aull, 0x3E1AF419A269A5F8ull, };
static struct ecc_curve gost_cp256b = {
.name = "cp256b",
.g = {
.x = cp256b_g_x,
.y = cp256b_g_y,
.ndigits = 256 / 64,
},
.p = cp256b_p,
.n = cp256b_n,
.a = cp256b_a,
.b = cp256b_b
};
/* OID_gostCPSignC 1.2.643.2.2.35.3 */
static u64 cp256c_g_x[] = {
0x0000000000000000ull, 0x0000000000000000ull,
0x0000000000000000ull, 0x0000000000000000ull, };
static u64 cp256c_g_y[] = {
0x366E550DFDB3BB67ull, 0x4D4DC440D4641A8Full,
0x3CBF3783CD08C0EEull, 0x41ECE55743711A8Cull, };
static u64 cp256c_p[] = {
0x7998F7B9022D759Bull, 0xCF846E86789051D3ull,
0xAB1EC85E6B41C8AAull, 0x9B9F605F5A858107ull,
/* pre-computed value for Barrett's reduction */
0xedc283cdd217b5a2ull, 0xbac48fc06398ae59ull,
0x405384d55f9f3b73ull, 0xa51f176161f1d734ull,
0x0000000000000001ull, };
static u64 cp256c_n[] = {
0xF02F3A6598980BB9ull, 0x582CA3511EDDFB74ull,
0xAB1EC85E6B41C8AAull, 0x9B9F605F5A858107ull, };
static u64 cp256c_a[] = { /* a = p - 3 */
0x7998F7B9022D7598ull, 0xCF846E86789051D3ull,
0xAB1EC85E6B41C8AAull, 0x9B9F605F5A858107ull, };
static u64 cp256c_b[] = {
0x000000000000805aull, 0x0000000000000000ull,
0x0000000000000000ull, 0x0000000000000000ull, };
static struct ecc_curve gost_cp256c = {
.name = "cp256c",
.g = {
.x = cp256c_g_x,
.y = cp256c_g_y,
.ndigits = 256 / 64,
},
.p = cp256c_p,
.n = cp256c_n,
.a = cp256c_a,
.b = cp256c_b
};
/* tc512{a,b} curves first recommended in 2013 and then standardized in
* R 50.1.114-2016 and RFC 7836 for use with GOST R 34.10-2012 (as TC26
* 512-bit ParamSet{A,B}).
*/
/* OID_gostTC26Sign512A 1.2.643.7.1.2.1.2.1 */
static u64 tc512a_g_x[] = {
0x0000000000000003ull, 0x0000000000000000ull,
0x0000000000000000ull, 0x0000000000000000ull,
0x0000000000000000ull, 0x0000000000000000ull,
0x0000000000000000ull, 0x0000000000000000ull, };
static u64 tc512a_g_y[] = {
0x89A589CB5215F2A4ull, 0x8028FE5FC235F5B8ull,
0x3D75E6A50E3A41E9ull, 0xDF1626BE4FD036E9ull,
0x778064FDCBEFA921ull, 0xCE5E1C93ACF1ABC1ull,
0xA61B8816E25450E6ull, 0x7503CFE87A836AE3ull, };
static u64 tc512a_p[] = { /* p = 2^512 - 569 */
0xFFFFFFFFFFFFFDC7ull, 0xFFFFFFFFFFFFFFFFull,
0xFFFFFFFFFFFFFFFFull, 0xFFFFFFFFFFFFFFFFull,
0xFFFFFFFFFFFFFFFFull, 0xFFFFFFFFFFFFFFFFull,
0xFFFFFFFFFFFFFFFFull, 0xFFFFFFFFFFFFFFFFull, };
static u64 tc512a_n[] = {
0xCACDB1411F10B275ull, 0x9B4B38ABFAD2B85Dull,
0x6FF22B8D4E056060ull, 0x27E69532F48D8911ull,
0xFFFFFFFFFFFFFFFFull, 0xFFFFFFFFFFFFFFFFull,
0xFFFFFFFFFFFFFFFFull, 0xFFFFFFFFFFFFFFFFull, };
static u64 tc512a_a[] = { /* a = p - 3 */
0xFFFFFFFFFFFFFDC4ull, 0xFFFFFFFFFFFFFFFFull,
0xFFFFFFFFFFFFFFFFull, 0xFFFFFFFFFFFFFFFFull,
0xFFFFFFFFFFFFFFFFull, 0xFFFFFFFFFFFFFFFFull,
0xFFFFFFFFFFFFFFFFull, 0xFFFFFFFFFFFFFFFFull, };
static u64 tc512a_b[] = {
0x503190785A71C760ull, 0x862EF9D4EBEE4761ull,
0x4CB4574010DA90DDull, 0xEE3CB090F30D2761ull,
0x79BD081CFD0B6265ull, 0x34B82574761CB0E8ull,
0xC1BD0B2B6667F1DAull, 0xE8C2505DEDFC86DDull, };
static struct ecc_curve gost_tc512a = {
.name = "tc512a",
.g = {
.x = tc512a_g_x,
.y = tc512a_g_y,
.ndigits = 512 / 64,
},
.p = tc512a_p,
.n = tc512a_n,
.a = tc512a_a,
.b = tc512a_b
};
/* OID_gostTC26Sign512B 1.2.643.7.1.2.1.2.2 */
static u64 tc512b_g_x[] = {
0x0000000000000002ull, 0x0000000000000000ull,
0x0000000000000000ull, 0x0000000000000000ull,
0x0000000000000000ull, 0x0000000000000000ull,
0x0000000000000000ull, 0x0000000000000000ull, };
static u64 tc512b_g_y[] = {
0x7E21340780FE41BDull, 0x28041055F94CEEECull,
0x152CBCAAF8C03988ull, 0xDCB228FD1EDF4A39ull,
0xBE6DD9E6C8EC7335ull, 0x3C123B697578C213ull,
0x2C071E3647A8940Full, 0x1A8F7EDA389B094Cull, };
static u64 tc512b_p[] = { /* p = 2^511 + 111 */
0x000000000000006Full, 0x0000000000000000ull,
0x0000000000000000ull, 0x0000000000000000ull,
0x0000000000000000ull, 0x0000000000000000ull,
0x0000000000000000ull, 0x8000000000000000ull, };
static u64 tc512b_n[] = {
0xC6346C54374F25BDull, 0x8B996712101BEA0Eull,
0xACFDB77BD9D40CFAull, 0x49A1EC142565A545ull,
0x0000000000000001ull, 0x0000000000000000ull,
0x0000000000000000ull, 0x8000000000000000ull, };
static u64 tc512b_a[] = { /* a = p - 3 */
0x000000000000006Cull, 0x0000000000000000ull,
0x0000000000000000ull, 0x0000000000000000ull,
0x0000000000000000ull, 0x0000000000000000ull,
0x0000000000000000ull, 0x8000000000000000ull, };
static u64 tc512b_b[] = {
0xFB8CCBC7C5140116ull, 0x50F78BEE1FA3106Eull,
0x7F8B276FAD1AB69Cull, 0x3E965D2DB1416D21ull,
0xBF85DC806C4B289Full, 0xB97C7D614AF138BCull,
0x7E3E06CF6F5E2517ull, 0x687D1B459DC84145ull, };
static struct ecc_curve gost_tc512b = {
.name = "tc512b",
.g = {
.x = tc512b_g_x,
.y = tc512b_g_y,
.ndigits = 512 / 64,
},
.p = tc512b_p,
.n = tc512b_n,
.a = tc512b_a,
.b = tc512b_b
};
#endif
Computing file changes ...
