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
ftrace.c
// SPDX-License-Identifier: GPL-2.0-only
/*
* Copyright 2012 Google, Inc.
*/
#include <linux/kernel.h>
#include <linux/compiler.h>
#include <linux/irqflags.h>
#include <linux/percpu.h>
#include <linux/smp.h>
#include <linux/atomic.h>
#include <linux/types.h>
#include <linux/mutex.h>
#include <linux/ftrace.h>
#include <linux/fs.h>
#include <linux/debugfs.h>
#include <linux/err.h>
#include <linux/cache.h>
#include <linux/slab.h>
#include <asm/barrier.h>
#include "internal.h"
/* This doesn't need to be atomic: speed is chosen over correctness here. */
static u64 pstore_ftrace_stamp;
static void notrace pstore_ftrace_call(unsigned long ip,
unsigned long parent_ip,
struct ftrace_ops *op,
struct pt_regs *regs)
{
unsigned long flags;
struct pstore_ftrace_record rec = {};
struct pstore_record record = {
.type = PSTORE_TYPE_FTRACE,
.buf = (char *)&rec,
.size = sizeof(rec),
.psi = psinfo,
};
if (unlikely(oops_in_progress))
return;
local_irq_save(flags);
rec.ip = ip;
rec.parent_ip = parent_ip;
pstore_ftrace_write_timestamp(&rec, pstore_ftrace_stamp++);
pstore_ftrace_encode_cpu(&rec, raw_smp_processor_id());
psinfo->write(&record);
local_irq_restore(flags);
}
static struct ftrace_ops pstore_ftrace_ops __read_mostly = {
.func = pstore_ftrace_call,
};
static DEFINE_MUTEX(pstore_ftrace_lock);
static bool pstore_ftrace_enabled;
static ssize_t pstore_ftrace_knob_write(struct file *f, const char __user *buf,
size_t count, loff_t *ppos)
{
u8 on;
ssize_t ret;
ret = kstrtou8_from_user(buf, count, 2, &on);
if (ret)
return ret;
mutex_lock(&pstore_ftrace_lock);
if (!on ^ pstore_ftrace_enabled)
goto out;
if (on) {
ftrace_ops_set_global_filter(&pstore_ftrace_ops);
ret = register_ftrace_function(&pstore_ftrace_ops);
} else {
ret = unregister_ftrace_function(&pstore_ftrace_ops);
}
if (ret) {
pr_err("%s: unable to %sregister ftrace ops: %zd\n",
__func__, on ? "" : "un", ret);
goto err;
}
pstore_ftrace_enabled = on;
out:
ret = count;
err:
mutex_unlock(&pstore_ftrace_lock);
return ret;
}
static ssize_t pstore_ftrace_knob_read(struct file *f, char __user *buf,
size_t count, loff_t *ppos)
{
char val[] = { '0' + pstore_ftrace_enabled, '\n' };
return simple_read_from_buffer(buf, count, ppos, val, sizeof(val));
}
static const struct file_operations pstore_knob_fops = {
.open = simple_open,
.read = pstore_ftrace_knob_read,
.write = pstore_ftrace_knob_write,
};
static struct dentry *pstore_ftrace_dir;
void pstore_register_ftrace(void)
{
if (!psinfo->write)
return;
pstore_ftrace_dir = debugfs_create_dir("pstore", NULL);
debugfs_create_file("record_ftrace", 0600, pstore_ftrace_dir, NULL,
&pstore_knob_fops);
}
void pstore_unregister_ftrace(void)
{
mutex_lock(&pstore_ftrace_lock);
if (pstore_ftrace_enabled) {
unregister_ftrace_function(&pstore_ftrace_ops);
pstore_ftrace_enabled = false;
}
mutex_unlock(&pstore_ftrace_lock);
debugfs_remove_recursive(pstore_ftrace_dir);
}
ssize_t pstore_ftrace_combine_log(char **dest_log, size_t *dest_log_size,
const char *src_log, size_t src_log_size)
{
size_t dest_size, src_size, total, dest_off, src_off;
size_t dest_idx = 0, src_idx = 0, merged_idx = 0;
void *merged_buf;
struct pstore_ftrace_record *drec, *srec, *mrec;
size_t record_size = sizeof(struct pstore_ftrace_record);
dest_off = *dest_log_size % record_size;
dest_size = *dest_log_size - dest_off;
src_off = src_log_size % record_size;
src_size = src_log_size - src_off;
total = dest_size + src_size;
merged_buf = kmalloc(total, GFP_KERNEL);
if (!merged_buf)
return -ENOMEM;
drec = (struct pstore_ftrace_record *)(*dest_log + dest_off);
srec = (struct pstore_ftrace_record *)(src_log + src_off);
mrec = (struct pstore_ftrace_record *)(merged_buf);
while (dest_size > 0 && src_size > 0) {
if (pstore_ftrace_read_timestamp(&drec[dest_idx]) <
pstore_ftrace_read_timestamp(&srec[src_idx])) {
mrec[merged_idx++] = drec[dest_idx++];
dest_size -= record_size;
} else {
mrec[merged_idx++] = srec[src_idx++];
src_size -= record_size;
}
}
while (dest_size > 0) {
mrec[merged_idx++] = drec[dest_idx++];
dest_size -= record_size;
}
while (src_size > 0) {
mrec[merged_idx++] = srec[src_idx++];
src_size -= record_size;
}
kfree(*dest_log);
*dest_log = merged_buf;
*dest_log_size = total;
return 0;
}
EXPORT_SYMBOL_GPL(pstore_ftrace_combine_log);
![swh spinner](/static/img/swh-spinner.gif)
Computing file changes ...