freebsd-dev/sys/kern/kern_vnodedumper.c
Mark Johnston c3179891f8 kerneldump: Inline dump_savectx() into its callers
The callers of dump_savectx() (i.e., doadump() and livedump_start())
subsequently call dumpsys()/minidumpsys(), which dump the calling
thread's stack when writing the dump.  If dump_savectx() gets its own
stack frame, that frame might be clobbered when its caller later calls
dumpsys()/minidumpsys(), making it difficult for debuggers to unwind the
stack.

Fix this by making dump_savectx() a macro, so that savectx() is always
called directly by the function which subsequently calls
dumpsys()/minidumpsys().

This fixes stack unwinding for the panicking thread from arm64
minidumps.  The same happened to work on amd64, but kgdb reports the
dump_savectx() calls as coming from dumpsys(), so in that case it
appears to work by accident.

Fixes:	c9114f9f86 ("Add new vnode dumper to support live minidumps")
Reviewed by:	mhorne, jhb
MFC after:	3 days
Differential Revision:	https://reviews.freebsd.org/D39151
2023-03-20 14:16:28 -04:00

213 lines
5.5 KiB
C

/*-
* Copyright (c) 2021-2022 Juniper Networks
*
* This software was developed by Mitchell Horne <mhorne@FreeBSD.org>
* under sponsorship from Juniper Networks and Klara Systems.
*
* Redistribution and use in source and binary forms, with or without
* modification, are permitted provided that the following conditions
* are met:
*
* 1. Redistributions of source code must retain the above copyright
* notice, this list of conditions and the following disclaimer.
* 2. Redistributions in binary form must reproduce the above copyright
* notice, this list of conditions and the following disclaimer in the
* documentation and/or other materials provided with the distribution.
*
* THIS SOFTWARE IS PROVIDED BY THE AUTHOR ``AS IS'' AND ANY EXPRESS OR
* IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE IMPLIED WARRANTIES
* OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE ARE DISCLAIMED.
* IN NO EVENT SHALL THE AUTHOR BE LIABLE FOR ANY DIRECT, INDIRECT,
* INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT
* NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE,
* DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY
* THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT
* (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE OF
* THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
*/
#include <sys/param.h>
#include <sys/systm.h>
#include <sys/conf.h>
#include <sys/caprights.h>
#include <sys/disk.h>
#include <sys/eventhandler.h>
#include <sys/fcntl.h>
#include <sys/file.h>
#include <sys/kerneldump.h>
#include <sys/limits.h>
#include <sys/malloc.h>
#include <sys/namei.h>
#include <sys/priv.h>
#include <sys/proc.h>
#include <sys/stat.h>
#include <sys/sysctl.h>
#include <sys/vnode.h>
#include <machine/pcb.h>
#include <machine/vmparam.h>
static dumper_start_t vnode_dumper_start;
static dumper_t vnode_dump;
static dumper_hdr_t vnode_write_headers;
static struct sx livedump_sx;
SX_SYSINIT(livedump, &livedump_sx, "Livedump sx");
/*
* Invoke a live minidump on the system.
*/
int
livedump_start(int fd, int flags, uint8_t compression)
{
#if MINIDUMP_PAGE_TRACKING == 1
struct dumperinfo di, *livedi;
struct diocskerneldump_arg kda;
struct vnode *vp;
struct file *fp;
void *rl_cookie;
int error;
error = priv_check(curthread, PRIV_KMEM_READ);
if (error != 0)
return (error);
if (flags != 0)
return (EINVAL);
error = getvnode(curthread, fd, &cap_write_rights, &fp);
if (error != 0)
return (error);
vp = fp->f_vnode;
if ((fp->f_flag & FWRITE) == 0) {
error = EBADF;
goto drop;
}
/* Set up a new dumper. */
bzero(&di, sizeof(di));
di.dumper_start = vnode_dumper_start;
di.dumper = vnode_dump;
di.dumper_hdr = vnode_write_headers;
di.blocksize = PAGE_SIZE; /* Arbitrary. */
di.maxiosize = MAXDUMPPGS * PAGE_SIZE;
bzero(&kda, sizeof(kda));
kda.kda_compression = compression;
error = dumper_create(&di, "livedump", &kda, &livedi);
if (error != 0)
goto drop;
/* Only allow one livedump to proceed at a time. */
if (sx_try_xlock(&livedump_sx) == 0) {
dumper_destroy(livedi);
error = EBUSY;
goto drop;
}
/* To be used by the callback functions. */
livedi->priv = vp;
/* Lock the entire file range and vnode. */
rl_cookie = vn_rangelock_wlock(vp, 0, OFF_MAX);
vn_lock(vp, LK_EXCLUSIVE | LK_RETRY);
EVENTHANDLER_INVOKE(livedumper_start, &error);
if (error != 0)
goto out;
dump_savectx();
error = minidumpsys(livedi, true);
EVENTHANDLER_INVOKE(livedumper_finish);
out:
VOP_UNLOCK(vp);
vn_rangelock_unlock(vp, rl_cookie);
sx_xunlock(&livedump_sx);
dumper_destroy(livedi);
drop:
fdrop(fp, curthread);
return (error);
#else
return (EOPNOTSUPP);
#endif /* MINIDUMP_PAGE_TRACKING == 1 */
}
int
vnode_dumper_start(struct dumperinfo *di, void *key, uint32_t keysize)
{
/* Always begin with an offset of zero. */
di->dumpoff = 0;
KASSERT(keysize == 0, ("encryption not supported for livedumps"));
return (0);
}
/*
* Callback from dumpsys() to dump a chunk of memory.
*
* Parameters:
* arg Opaque private pointer to vnode
* virtual Virtual address (where to read the data from)
* offset Offset from start of core file
* length Data length
*
* Return value:
* 0 on success
* errno on error
*/
int
vnode_dump(void *arg, void *virtual, off_t offset, size_t length)
{
struct vnode *vp;
int error = 0;
vp = arg;
MPASS(vp != NULL);
ASSERT_VOP_LOCKED(vp, __func__);
EVENTHANDLER_INVOKE(livedumper_dump, virtual, offset, length, &error);
if (error != 0)
return (error);
/* Done? */
if (virtual == NULL)
return (0);
error = vn_rdwr(UIO_WRITE, vp, virtual, length, offset, UIO_SYSSPACE,
IO_NODELOCKED, curthread->td_ucred, NOCRED, NULL, curthread);
if (error != 0)
uprintf("%s: error writing livedump block at offset %jx: %d\n",
__func__, (uintmax_t)offset, error);
return (error);
}
/*
* Callback from dumpsys() to write out the dump header, placed at the end.
*/
int
vnode_write_headers(struct dumperinfo *di, struct kerneldumpheader *kdh)
{
struct vnode *vp;
int error;
off_t offset;
vp = di->priv;
MPASS(vp != NULL);
ASSERT_VOP_LOCKED(vp, __func__);
/* Compensate for compression/encryption adjustment of dumpoff. */
offset = roundup2(di->dumpoff, di->blocksize);
/* Write the kernel dump header to the end of the file. */
error = vn_rdwr(UIO_WRITE, vp, kdh, sizeof(*kdh), offset,
UIO_SYSSPACE, IO_NODELOCKED, curthread->td_ucred, NOCRED, NULL,
curthread);
if (error != 0)
uprintf("%s: error writing livedump header: %d\n", __func__,
error);
return (error);
}