Bug 260012 - NFS v4 client can crash server with a bad LAYOUTRETURN RPC
Summary: NFS v4 client can crash server with a bad LAYOUTRETURN RPC
Status: Open
Alias: None
Product: Base System
Classification: Unclassified
Component: kern (show other bugs)
Version: CURRENT
Hardware: Any Any
: --- Affects Some People
Assignee: Rick Macklem
URL:
Keywords:
Depends on:
Blocks:
 
Reported: 2021-11-24 09:47 UTC by Robert Morris
Modified: 2021-11-26 16:45 UTC (History)
2 users (show)

See Also:


Attachments
An NFS client that crashes the server with a bad LAYOUTRETURN. (29.99 KB, text/plain)
2021-11-24 09:47 UTC, Robert Morris
no flags Details
Check for a non-NULL layp before calling nfsrv_flexlayouterr() (1.04 KB, patch)
2021-11-25 23:58 UTC, Rick Macklem
no flags Details | Diff

Note You need to log in before you can comment on or make changes to this bug.
Description Robert Morris 2021-11-24 09:47:47 UTC
Created attachment 229690 [details]
An NFS client that crashes the server with a bad LAYOUTRETURN.

If an NFS v4 client sends a LAYOUTRETURN RPC with lr_returntype of
NFSV4LAYOURRET_FILE and an lrf_body<> size <= 0, nfsrvd_layoutreturn()
will pass a NULL layp to nfsrv_layoutreturn(). If the RPC also has
layouttype==NFSV4LAYOUT_FLEXFILE (4), nfsrv_layoutreturn() will call
nfsrv_flexlayouterr(). The latter dereferences layp (as tl) without
checking for NULL.

I've attached a demo:

# uname -a
FreeBSD  14.0-CURRENT FreeBSD 14.0-CURRENT #96 main-n250901-77e3db078984-dirty: Tue Nov 23 16:58:28 EST 2021     rtm@xxx:/usr/obj/usr/rtm/symbsd/src/riscv.riscv64/sys/RTM  riscv
# cc fnfsd_1.c
# ./a.out
...
nfsrv_layoutreturn: updatemdsattr failed=2
panic: Fatal page fault at 0xffffffc00025ce8e: 0000000000000000
cpuid = 0
time = 1637591542
KDB: stack backtrace:
db_trace_self() at db_trace_self
db_trace_self_wrapper() at db_trace_self_wrapper+0x38
kdb_backtrace() at kdb_backtrace+0x2c
vpanic() at vpanic+0x154
panic() at panic+0x2a
page_fault_handler() at page_fault_handler+0x1ee
do_trap_supervisor() at do_trap_supervisor+0x76
cpu_exception_handler_supervisor() at cpu_exception_handler_supervisor+0x70
--- exception 13, tval = 0
nfsrv_layoutreturn() at nfsrv_layoutreturn+0xa6
nfsrvd_layoutreturn() at nfsrvd_layoutreturn+0x35c
nfsrvd_dorpc() at nfsrvd_dorpc+0x154c
nfssvc_program() at nfssvc_program+0x5a4
svc_run_internal() at svc_run_internal+0x808
svc_run() at svc_run+0x1a8
nfsrvd_nfsd() at nfsrvd_nfsd+0x30e
nfssvc_nfsd() at nfssvc_nfsd+0x386
sys_nfssvc() at sys_nfssvc+0xd0
do_trap_user() at do_trap_user+0x206
cpu_exception_handler_user() at cpu_exception_handler_user+0x72
Comment 1 Rick Macklem freebsd_committer 2021-11-25 23:58:35 UTC
Created attachment 229733 [details]
Check for a non-NULL layp before calling nfsrv_flexlayouterr()

This patch should avoid the crash.
While here I also added a sanity check on the
size of the error data.

Maybe the reporter can test to see this fixes the problem?
Comment 2 Robert Morris 2021-11-26 16:45:15 UTC
(In reply to Rick Macklem from comment #1)
The patch fixes the crash for me -- thank you.