d/freebsd-dev
1
0
Fork 0
Code Issues Pull Requests Packages Projects Releases Wiki Activity

Document lock order reversals possible when calling allocation functions

Browse Source 
with vnode interlocks held.

Emphasize that users need to be careful with malloc flags versus mbuf
flags.

Remove stale portion in vnode.9 about v_tag.  We don't have it anylonger.

Submitted by: Hiten Pandya <hiten@unixdaemons.com>
This commit is contained in:
Alfred Perlstein 2003-01-18 19:51:52 +00:00
parent 314e5a3daf
commit 79525d59b2
Notes: svn2git 2020-12-20 02:59:44 +00:00 
svn path=/head/; revision=109493
2 changed files with 28 additions and 14 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

16
share/man/man9/malloc.9
View File

@ -240,8 +240,24 @@ of two for requests up to the size of a page of memory.
For larger requests, one or more pages is allocated.
While it should not be relied upon, this information may be useful for
optimizing the efficiency of memory use.
.Pp
Malloc flags documented above should
.Em NOT
be used with
.Xr mbuf 9
routines as it will cause undesired results.
.Pp
Any calls to
.Fn malloc
or
.Fn free
when holding a
.Xr vnode 9
interlock, will cause a LOR (Lock Order Reversal) due to the
interwining of VM Objects and Vnodes.
.Sh SEE ALSO
.Xr vmstat 8
.Xr vnode 9
.Sh DIAGNOSTICS
A kernel compiled with the
.Dv DIAGNOSTIC

26
share/man/man9/vnode.9
View File

@ -44,18 +44,6 @@
*/
enum vtype { VNON, VREG, VDIR, VBLK, VCHR, VLNK, VSOCK, VFIFO, VBAD };
/*
* Vnode tag types.
* These are for the benefit of external programs only (e.g., pstat)
* and should NEVER be inspected by the kernel.
*/
enum vtagtype {
VT_NON, VT_UFS, VT_NFS, VT_UNUSED, VT_PC, VT_LFS, VT_LOFS, VT_FDESC,
VT_PORTAL, VT_NULL, VT_UMAP, VT_KERNFS, VT_PROCFS, VT_AFS, VT_ISOFS,
VT_UNION, VT_MSDOSFS, VT_DEVFS, VT_TFS, VT_VFS, VT_CODA, VT_NTFS,
VT_HPFS, VT_NWFS, VT_PSEUDOFS
};
/*
* Each underlying file system allocates its own private area and hangs
* it from v_data. If non-null, this area is freed in getnewvnode().
@ -105,7 +93,7 @@ struct vnode {
struct vm_object *v_object; /* Place to store VM object */
struct simplelock v_interlock; /* lock on usecount and flag */
struct lock *v_vnlock; /* used for non-locking fs's */
enum vtagtype v_tag; /* type of underlying data */
const char *v_tag; /* type of underlying data */
void *v_data; /* private data for fs */
LIST_HEAD(, namecache) v_cache_src; /* Cache entries from us */
TAILQ_HEAD(, namecache) v_cache_dst; /* Cache entries to us */
@ -241,7 +229,7 @@ A FIFO (named pipe). Advisory locking won't work on this.
.It Dv VBAD
An old style bad sector map
.El
.Sh NOTES
.Sh IMPLEMENTATION NOTES
VFIFO uses the "struct fileops" from
.Pa /sys/kern/sys_pipe.c .
VSOCK uses the "struct fileops" from
@ -252,7 +240,17 @@ Everything else uses the one from
The VFIFO/VSOCK code, which is why "struct fileops" is used at all, is
an artifact of an incomplete integration of the VFS code into the
kernel.
.Pp
Calls to
.Xr malloc 9
or
.Xr free 9
when holding a
.Nm
interlock, will cause a LOR (Lock Order Reversal) due to the
interwining of VM Objects and Vnodes.
.Sh SEE ALSO
.Xr malloc 9
.Xr VFS 9
.Sh AUTHORS
This man page was written by