Hi David,
I noticed 2 other issues while running xfstests.
Noticed kernel OOPS during test generic/286:
folio_test_writeback returned false which means PG_writeback flag has
been cleared. I am not sure whether head page has PG_writeback flag
set initially? Can you please confirm.
[ 2275.941096] CIFS: bad 2000 @64f0000 page 64f0 64f1
[ 2275.945785] ------------[ cut here ]------------
[ 2275.945787] kernel BUG at
/home/lxsmbadmin/latest_14mar/linux-fs/fs/cifs/cifssmb.c:1954!
[ 2275.952198] invalid opcode: 0000 [#1] PREEMPT SMP PTI
[ 2275.956500] CPU: 0 PID: 3369 Comm: kworker/0:7 Tainted: G
OE 5.17.0-rc6+ #1
[ 2275.962573] Hardware name: Microsoft Corporation Virtual
Machine/Virtual Machine, BIOS Hyper-V UEFI Release v4.1 10/27/2020
[ 2275.969812] Workqueue: cifsiod cifs_writev_complete [cifs]
[ 2275.974909] RIP: 0010:cifs_pages_written_back+0x1e1/0x1f0 [cifs]
[ 2275.975570] CIFS: bad 2000 @64f0000 page 64f0 64f1
[ 2275.980619] Code: 00 48 8b 07 f6 c4 04 0f 84 92 84 06 00 e8 77 a6
e6 db 48 89 c1 49 89 d8 4c 89 e2 44 89 ee 48 c7 c7 e0 f5 8e c0 e8 a4
9d 71 dc <0f> 0b e8 58 f5 76 dc 0f 1f 84 00 00 00 00 00 0f 1f 44 00 00
55 48
[ 2275.980622] RSP: 0018:ffffac614177fdc8 EFLAGS: 00010246
[ 2275.980625] RAX: 0000000000000026 RBX: 00000000000064f1 RCX: 0000000000000000
[ 2275.980626] RDX: 0000000000000000 RSI: ffffffff9d767af1 RDI: 00000000ffffffff
[ 2275.980628] RBP: ffffac614177fe28 R08: 0000000000000000 R09: ffffac614177fbc0
[ 2275.980629] R10: ffffac614177fbb8 R11: ffffffff9df52b68 R12: 00000000064f0000
[ 2275.980631] R13: 0000000000002000 R14: 0000000000000000 R15: ffff9c8193bc7e40
[ 2275.980632] FS: 0000000000000000(0000) GS:ffff9c8337c00000(0000)
knlGS:0000000000000000
[ 2275.980636] CS: 0010 DS: 0000 ES: 0000 CR0: 0000000080050033
[ 2275.980638] CR2: 00007fd099e56b90 CR3: 00000001028e4006 CR4: 00000000003706f0
[ 2275.980639] DR0: 0000000000000000 DR1: 0000000000000000 DR2: 0000000000000000
[ 2275.980640] DR3: 0000000000000000 DR6: 00000000fffe0ff0 DR7: 0000000000000400
[ 2275.980641] Call Trace:
[ 2275.980643] <TASK>
[ 2275.980648] cifs_writev_complete+0x43d/0x500 [cifs]
Noticed that with netfs integration, file open with O_DIRECT flag is
not supported.
Regards,
Rohith
On Tue, Mar 15, 2022 at 9:20 AM Rohith Surabattula
<rohiths.msft@xxxxxxxxx> wrote:
>
> Hi David,
>
> Below change needs to be applied on top of your branch.
>
> lxsmbadmin@netfsvm:~/latest_14mar/linux-fs$ git diff
> diff --git a/fs/cifs/file.c b/fs/cifs/file.c
> index af7483c246ac..447934ff80b8 100644
> --- a/fs/cifs/file.c
> +++ b/fs/cifs/file.c
> @@ -2969,6 +2969,12 @@ cifs_write_from_iter(loff_t offset, size_t len,
> struct iov_iter *from,
>
> cur_len = min_t(const size_t, len, wsize);
>
> + if (!cur_len) {
> + rc = -EAGAIN;
> + add_credits_and_wake_if(server, credits, 0);
> + break;
> + }
> +
> wdata = cifs_writedata_alloc(cifs_uncached_writev_complete);
> if (!wdata) {
> rc = -ENOMEM;
>
> lxsmbadmin@netfsvm:~/xfstests-dev$ sudo ./check generic/013
> SECTION -- smb3
> FSTYP -- cifs
> PLATFORM -- Linux/x86_64 netfsvm 5.17.0-rc6+ #1 SMP PREEMPT Mon
> Mar 14 09:05:47 UTC 2022
> MKFS_OPTIONS -- //127.0.0.1/sambashare_scratch
> MOUNT_OPTIONS -- -ousername=,password=,noperm,vers=3.0,actimeo=0
> //127.0.0.1/sambashare_scratch /mnt/xfstests_scratch
>
> generic/013 149s
> Ran: generic/013
> Passed all 1 tests
>
> SECTION -- smb3
> =========================
> Ran: generic/013
> Passed all 1 tests
>
> Regards,
> Rohith
>
> On Fri, Mar 11, 2022 at 1:56 PM David Howells <dhowells@xxxxxxxxxx> wrote:
> >
> >
> > David Howells <dhowells@xxxxxxxxxx> wrote:
> >
> > > The other issue is that if I run splice to an empty file, it works; running
> > > another splice to the same file will result in the server giving
> > > STATUS_ACCESS_DENIED when cifs_write_begin() tries to read from the file:
> > >
> > > 7 0.009485249 192.168.6.2 → 192.168.6.1 SMB2 183 Read Request Len:65536 Off:0 File: x
> > > 8 0.009674245 192.168.6.1 → 192.168.6.2 SMB2 143 Read Response, Error: STATUS_ACCESS_DENIED
> > >
> > > Actually - that might be because the file is only 65536 bytes long because the
> > > first splice finished short.
> >
> > Actually, it's because I opened the output file O_WRONLY. If I open it
> > O_RDWR, it works. The test program is attached below.
> >
> > David
> > ---
> > #define _GNU_SOURCE
> > #include <stdio.h>
> > #include <stdlib.h>
> > #include <string.h>
> > #include <unistd.h>
> > #include <fcntl.h>
> >
> > int main(int argc, char *argv[])
> > {
> > off64_t opos;
> > size_t len;
> > int in, out;
> >
> > if (argc != 4) {
> > printf("Format: %s size in out\n", argv[0]);
> > exit(2);
> > }
> >
> > len = atol(argv[1]);
> >
> > if (strcmp(argv[2], "-") != 0) {
> > in = open(argv[2], O_RDONLY);
> > if (in < 0) {
> > perror(argv[2]);
> > return 1;
> > }
> > } else {
> > in = 0;
> > }
> >
> > if (strcmp(argv[3], "-") != 0) {
> > out = open(argv[3], O_WRONLY); // Change to O_RDWR
> > if (out < 0) {
> > perror(argv[3]);
> > return 1;
> > }
> > } else {
> > out = 1;
> > }
> >
> > opos = 3;
> > if (splice(in, NULL, out, &opos, len, 0) < 0) {
> > perror("splice");
> > return 1;
> > }
> >
> > if (close(in) < 0) {
> > perror("close/in");
> > return 1;
> > }
> >
> > if (close(out) < 0) {
> > perror("close/out");
> > return 1;
> > }
> >
> > return 0;
> > }
> >
