[U-Boot] NFS time out problem

older
[U-Boot] [PATCH 00/10] PXE support...

Matthias Brugger

11 Dec 2012 11 Dec '12

7:14 p.m.

After reading through the RFC [1] I realized that we have to increment the rpc_id when we resend a message. This patch therefor doesn't change the rpc_id behaviour, but adds a dynamic increment of the "time to wait before time out". Apart we just drop messages which are replies to timed out requests.

Any comments are welcome.

Show replies by date

Matthias Brugger

11 Dec 11 Dec

7:14 p.m.

New subject: [U-Boot] [PATCH] net: nfs: add dynamic wait period

This patch tackles the time out problem which leads to break the boot process, when loading file over nfs. The patch does two things.

First of all, we just ignore messages that arrive with a rpc_id smaller then the client id. We just interpret this messages as answers to formaly timed out messages.

Second, when a time out occurs we double the time to wait, so that we do not stress the server resending the last message.

Signed-off-by: Matthias Brugger matthias.bgg@gmail.com --- net/nfs.c | 73 +++++++++++++++++++++++++++++++++++++++++++------------------ 1 file changed, 52 insertions(+), 21 deletions(-)

diff --git a/net/nfs.c b/net/nfs.c index 7f2393f..84aeda1 100644 --- a/net/nfs.c +++ b/net/nfs.c @@ -37,10 +37,14 @@ # define NFS_TIMEOUT CONFIG_NFS_TIMEOUT #endif

+#define NFS_RPC_ERR 1 +#define NFS_RPC_DROP 124 + static int fs_mounted; static unsigned long rpc_id; static int nfs_offset = -1; static int nfs_len; +static ulong nfs_timeout = NFS_TIMEOUT;

static char dirfh[NFS_FHSIZE]; /* file handle of directory */ static char filefh[NFS_FHSIZE]; /* file handle of kernel image */ @@ -399,8 +403,10 @@ rpc_lookup_reply(int prog, uchar *pkt, unsigned len)

debug("%s\n", __func__);

- if (ntohl(rpc_pkt.u.reply.id) != rpc_id) - return -1; + if (ntohl(rpc_pkt.u.reply.id) > rpc_id) + return -NFS_RPC_ERR; + else if (ntohl(rpc_pkt.u.reply.id) < rpc_id) + return -NFS_RPC_DROP;

if (rpc_pkt.u.reply.rstatus || rpc_pkt.u.reply.verifier || @@ -428,8 +434,10 @@ nfs_mount_reply(uchar *pkt, unsigned len)

memcpy((unsigned char *)&rpc_pkt, pkt, len);

- if (ntohl(rpc_pkt.u.reply.id) != rpc_id) - return -1; + if (ntohl(rpc_pkt.u.reply.id) > rpc_id) + return -NFS_RPC_ERR; + else if (ntohl(rpc_pkt.u.reply.id) < rpc_id) + return -NFS_RPC_DROP;

if (rpc_pkt.u.reply.rstatus || rpc_pkt.u.reply.verifier || @@ -452,8 +460,10 @@ nfs_umountall_reply(uchar *pkt, unsigned len)

memcpy((unsigned char *)&rpc_pkt, pkt, len);

- if (ntohl(rpc_pkt.u.reply.id) != rpc_id) - return -1; + if (ntohl(rpc_pkt.u.reply.id) > rpc_id) + return -NFS_RPC_ERR; + else if (ntohl(rpc_pkt.u.reply.id) < rpc_id) + return -NFS_RPC_DROP;

if (rpc_pkt.u.reply.rstatus || rpc_pkt.u.reply.verifier || @@ -475,8 +485,10 @@ nfs_lookup_reply(uchar *pkt, unsigned len)

memcpy((unsigned char *)&rpc_pkt, pkt, len);

- if (ntohl(rpc_pkt.u.reply.id) != rpc_id) - return -1; + if (ntohl(rpc_pkt.u.reply.id) > rpc_id) + return -NFS_RPC_ERR; + else if (ntohl(rpc_pkt.u.reply.id) < rpc_id) + return -NFS_RPC_DROP;

if (rpc_pkt.u.reply.rstatus || rpc_pkt.u.reply.verifier || @@ -499,8 +511,10 @@ nfs_readlink_reply(uchar *pkt, unsigned len)

memcpy((unsigned char *)&rpc_pkt, pkt, len);

- if (ntohl(rpc_pkt.u.reply.id) != rpc_id) - return -1; + if (ntohl(rpc_pkt.u.reply.id) > rpc_id) + return -NFS_RPC_ERR; + else if (ntohl(rpc_pkt.u.reply.id) < rpc_id) + return -NFS_RPC_DROP;

if (rpc_pkt.u.reply.rstatus || rpc_pkt.u.reply.verifier || @@ -534,8 +548,10 @@ nfs_read_reply(uchar *pkt, unsigned len)

memcpy((uchar *)&rpc_pkt, pkt, sizeof(rpc_pkt.u.reply));

- if (ntohl(rpc_pkt.u.reply.id) != rpc_id) - return -1; + if (ntohl(rpc_pkt.u.reply.id) > rpc_id) + return -NFS_RPC_ERR; + else if (ntohl(rpc_pkt.u.reply.id) < rpc_id) + return -NFS_RPC_DROP;

if (rpc_pkt.u.reply.rstatus || rpc_pkt.u.reply.verifier || @@ -574,7 +590,7 @@ NfsTimeout(void) NetStartAgain(); } else { puts("T "); - NetSetTimeout(NFS_TIMEOUT, NfsTimeout); + NetSetTimeout(nfs_timeout + NFS_TIMEOUT * NfsTimeoutCount, NfsTimeout); NfsSend(); } } @@ -583,6 +599,7 @@ static void NfsHandler(uchar *pkt, unsigned dest, IPaddr_t sip, unsigned src, unsigned len) { int rlen; + int reply;

debug("%s\n", __func__);

@@ -591,19 +608,24 @@ NfsHandler(uchar *pkt, unsigned dest, IPaddr_t sip, unsigned src, unsigned len)

switch (NfsState) { case STATE_PRCLOOKUP_PROG_MOUNT_REQ: - rpc_lookup_reply(PROG_MOUNT, pkt, len); + if (rpc_lookup_reply(PROG_MOUNT, pkt, len) == -NFS_RPC_DROP) + break; NfsState = STATE_PRCLOOKUP_PROG_NFS_REQ; NfsSend(); break;

case STATE_PRCLOOKUP_PROG_NFS_REQ: - rpc_lookup_reply(PROG_NFS, pkt, len); + if (rpc_lookup_reply(PROG_NFS, pkt, len) == -NFS_RPC_DROP) + break; NfsState = STATE_MOUNT_REQ; NfsSend(); break;

case STATE_MOUNT_REQ: - if (nfs_mount_reply(pkt, len)) { + reply = nfs_mount_reply(pkt, len); + if (reply == -NFS_RPC_DROP) + break; + else if (reply == -NFS_RPC_ERR) { puts("*** ERROR: Cannot mount\n"); /* just to be sure... */ NfsState = STATE_UMOUNT_REQ; @@ -615,7 +637,10 @@ NfsHandler(uchar *pkt, unsigned dest, IPaddr_t sip, unsigned src, unsigned len) break;

case STATE_UMOUNT_REQ: - if (nfs_umountall_reply(pkt, len)) { + reply = nfs_umountall_reply(pkt, len); + if (reply == -NFS_RPC_DROP) + break; + else if (reply == -NFS_RPC_ERR) { puts("*** ERROR: Cannot umount\n"); net_set_state(NETLOOP_FAIL); } else { @@ -625,7 +650,10 @@ NfsHandler(uchar *pkt, unsigned dest, IPaddr_t sip, unsigned src, unsigned len) break;

case STATE_LOOKUP_REQ: - if (nfs_lookup_reply(pkt, len)) { + reply = nfs_lookup_reply(pkt, len); + if (reply == -NFS_RPC_DROP) + break; + else if (reply == -NFS_RPC_ERR) { puts("*** ERROR: File lookup fail\n"); NfsState = STATE_UMOUNT_REQ; NfsSend(); @@ -638,7 +666,10 @@ NfsHandler(uchar *pkt, unsigned dest, IPaddr_t sip, unsigned src, unsigned len) break;

case STATE_READLINK_REQ: - if (nfs_readlink_reply(pkt, len)) { + reply = nfs_readlink_reply(pkt, len); + if (reply == -NFS_RPC_DROP) + break; + else if (reply == -NFS_RPC_ERR) { puts("*** ERROR: Symlink fail\n"); NfsState = STATE_UMOUNT_REQ; NfsSend(); @@ -654,7 +685,7 @@ NfsHandler(uchar *pkt, unsigned dest, IPaddr_t sip, unsigned src, unsigned len)

case STATE_READ_REQ: rlen = nfs_read_reply(pkt, len); - NetSetTimeout(NFS_TIMEOUT, NfsTimeout); + NetSetTimeout(nfs_timeout, NfsTimeout); if (rlen > 0) { nfs_offset += rlen; NfsSend(); @@ -738,7 +769,7 @@ NfsStart(void) printf("\nLoad address: 0x%lx\n" "Loading: *\b", load_addr);

- NetSetTimeout(NFS_TIMEOUT, NfsTimeout); + NetSetTimeout(nfs_timeout, NfsTimeout); net_set_udp_handler(NfsHandler);

NfsTimeoutCount = 0;

-- 1.7.9.5

Matthias Brugger

26 Jan 26 Jan

4:23 p.m.

New subject: [U-Boot] [PATCH] net: nfs: add dynamic wait period

2012/12/11 Matthias Brugger matthias.bgg@gmail.com:

...

This patch tackles the time out problem which leads to break the boot process, when loading file over nfs. The patch does two things.

First of all, we just ignore messages that arrive with a rpc_id smaller then the client id. We just interpret this messages as answers to formaly timed out messages.

Second, when a time out occurs we double the time to wait, so that we do not stress the server resending the last message.

Any comment on the patch?

...

Signed-off-by: Matthias Brugger matthias.bgg@gmail.com

net/nfs.c | 73 +++++++++++++++++++++++++++++++++++++++++++------------------ 1 file changed, 52 insertions(+), 21 deletions(-)

diff --git a/net/nfs.c b/net/nfs.c index 7f2393f..84aeda1 100644 --- a/net/nfs.c +++ b/net/nfs.c @@ -37,10 +37,14 @@ # define NFS_TIMEOUT CONFIG_NFS_TIMEOUT #endif

+#define NFS_RPC_ERR 1 +#define NFS_RPC_DROP 124

static int fs_mounted; static unsigned long rpc_id; static int nfs_offset = -1; static int nfs_len; +static ulong nfs_timeout = NFS_TIMEOUT;

    debug("%s\n", __func__);

  if (ntohl(rpc_pkt.u.reply.id) != rpc_id)

```
          return -1;
```

  if (ntohl(rpc_pkt.u.reply.id) > rpc_id)

```
          return -NFS_RPC_ERR;
```

  else if (ntohl(rpc_pkt.u.reply.id) < rpc_id)

          return -NFS_RPC_DROP;

  if (rpc_pkt.u.reply.rstatus  ||
      rpc_pkt.u.reply.verifier ||

@@ -428,8 +434,10 @@ nfs_mount_reply(uchar *pkt, unsigned len)

    memcpy((unsigned char *)&rpc_pkt, pkt, len);

  if (ntohl(rpc_pkt.u.reply.id) != rpc_id)

```
          return -1;
```

  if (ntohl(rpc_pkt.u.reply.id) > rpc_id)

```
          return -NFS_RPC_ERR;
```

  else if (ntohl(rpc_pkt.u.reply.id) < rpc_id)

          return -NFS_RPC_DROP;

  if (rpc_pkt.u.reply.rstatus  ||
      rpc_pkt.u.reply.verifier ||

@@ -452,8 +460,10 @@ nfs_umountall_reply(uchar *pkt, unsigned len)

    memcpy((unsigned char *)&rpc_pkt, pkt, len);

  if (ntohl(rpc_pkt.u.reply.id) != rpc_id)

```
          return -1;
```

  if (ntohl(rpc_pkt.u.reply.id) > rpc_id)

```
          return -NFS_RPC_ERR;
```

  else if (ntohl(rpc_pkt.u.reply.id) < rpc_id)

          return -NFS_RPC_DROP;

  if (rpc_pkt.u.reply.rstatus  ||
      rpc_pkt.u.reply.verifier ||

@@ -475,8 +485,10 @@ nfs_lookup_reply(uchar *pkt, unsigned len)

    memcpy((unsigned char *)&rpc_pkt, pkt, len);

  if (ntohl(rpc_pkt.u.reply.id) != rpc_id)

```
          return -1;
```

  if (ntohl(rpc_pkt.u.reply.id) > rpc_id)

```
          return -NFS_RPC_ERR;
```

  else if (ntohl(rpc_pkt.u.reply.id) < rpc_id)

          return -NFS_RPC_DROP;

  if (rpc_pkt.u.reply.rstatus  ||
      rpc_pkt.u.reply.verifier ||

@@ -499,8 +511,10 @@ nfs_readlink_reply(uchar *pkt, unsigned len)

    memcpy((unsigned char *)&rpc_pkt, pkt, len);

  if (ntohl(rpc_pkt.u.reply.id) != rpc_id)

```
          return -1;
```

  if (ntohl(rpc_pkt.u.reply.id) > rpc_id)

```
          return -NFS_RPC_ERR;
```

  else if (ntohl(rpc_pkt.u.reply.id) < rpc_id)

          return -NFS_RPC_DROP;

  if (rpc_pkt.u.reply.rstatus  ||
      rpc_pkt.u.reply.verifier ||

@@ -534,8 +548,10 @@ nfs_read_reply(uchar *pkt, unsigned len)

    memcpy((uchar *)&rpc_pkt, pkt, sizeof(rpc_pkt.u.reply));

  if (ntohl(rpc_pkt.u.reply.id) != rpc_id)

```
          return -1;
```

  if (ntohl(rpc_pkt.u.reply.id) > rpc_id)

```
          return -NFS_RPC_ERR;
```

  else if (ntohl(rpc_pkt.u.reply.id) < rpc_id)

          return -NFS_RPC_DROP;

  if (rpc_pkt.u.reply.rstatus  ||
      rpc_pkt.u.reply.verifier ||

@@ -574,7 +590,7 @@ NfsTimeout(void) NetStartAgain(); } else { puts("T ");

          NetSetTimeout(NFS_TIMEOUT, NfsTimeout);

          NetSetTimeout(nfs_timeout + NFS_TIMEOUT * NfsTimeoutCount, NfsTimeout);
          NfsSend();
  }

} @@ -583,6 +599,7 @@ static void NfsHandler(uchar *pkt, unsigned dest, IPaddr_t sip, unsigned src, unsigned len) { int rlen;

  int reply;

  debug("%s\n", __func__);

@@ -591,19 +608,24 @@ NfsHandler(uchar *pkt, unsigned dest, IPaddr_t sip, unsigned src, unsigned len)

    switch (NfsState) {
    case STATE_PRCLOOKUP_PROG_MOUNT_REQ:

          rpc_lookup_reply(PROG_MOUNT, pkt, len);

          if (rpc_lookup_reply(PROG_MOUNT, pkt, len) == -NFS_RPC_DROP)

                  break;
          NfsState = STATE_PRCLOOKUP_PROG_NFS_REQ;
          NfsSend();
          break;

  case STATE_PRCLOOKUP_PROG_NFS_REQ:

          rpc_lookup_reply(PROG_NFS, pkt, len);

          if (rpc_lookup_reply(PROG_NFS, pkt, len) == -NFS_RPC_DROP)

                  break;
          NfsState = STATE_MOUNT_REQ;
          NfsSend();
          break;

  case STATE_MOUNT_REQ:

          if (nfs_mount_reply(pkt, len)) {

          reply = nfs_mount_reply(pkt, len);

```
          if (reply == -NFS_RPC_DROP)
```
```
                  break;
```

          else if (reply == -NFS_RPC_ERR) {
                  puts("*** ERROR: Cannot mount\n");
                  /* just to be sure... */
                  NfsState = STATE_UMOUNT_REQ;

@@ -615,7 +637,10 @@ NfsHandler(uchar *pkt, unsigned dest, IPaddr_t sip, unsigned src, unsigned len) break;

    case STATE_UMOUNT_REQ:

          if (nfs_umountall_reply(pkt, len)) {

          reply = nfs_umountall_reply(pkt, len);

```
          if (reply == -NFS_RPC_DROP)
```
```
                  break;
```

          else if (reply == -NFS_RPC_ERR) {
                  puts("*** ERROR: Cannot umount\n");
                  net_set_state(NETLOOP_FAIL);
          } else {

@@ -625,7 +650,10 @@ NfsHandler(uchar *pkt, unsigned dest, IPaddr_t sip, unsigned src, unsigned len) break;

    case STATE_LOOKUP_REQ:

          if (nfs_lookup_reply(pkt, len)) {

          reply = nfs_lookup_reply(pkt, len);

```
          if (reply == -NFS_RPC_DROP)
```
```
                  break;
```

          else if (reply == -NFS_RPC_ERR) {
                  puts("*** ERROR: File lookup fail\n");
                  NfsState = STATE_UMOUNT_REQ;
                  NfsSend();

@@ -638,7 +666,10 @@ NfsHandler(uchar *pkt, unsigned dest, IPaddr_t sip, unsigned src, unsigned len) break;

    case STATE_READLINK_REQ:

          if (nfs_readlink_reply(pkt, len)) {

          reply = nfs_readlink_reply(pkt, len);

```
          if (reply == -NFS_RPC_DROP)
```
```
                  break;
```

          else if (reply == -NFS_RPC_ERR) {
                  puts("*** ERROR: Symlink fail\n");
                  NfsState = STATE_UMOUNT_REQ;
                  NfsSend();

@@ -654,7 +685,7 @@ NfsHandler(uchar *pkt, unsigned dest, IPaddr_t sip, unsigned src, unsigned len)

    case STATE_READ_REQ:
            rlen = nfs_read_reply(pkt, len);

          NetSetTimeout(NFS_TIMEOUT, NfsTimeout);

          NetSetTimeout(nfs_timeout, NfsTimeout);
          if (rlen > 0) {
                  nfs_offset += rlen;
                  NfsSend();

@@ -738,7 +769,7 @@ NfsStart(void) printf("\nLoad address: 0x%lx\n" "Loading: *\b", load_addr);

  NetSetTimeout(NFS_TIMEOUT, NfsTimeout);

  NetSetTimeout(nfs_timeout, NfsTimeout);
  net_set_udp_handler(NfsHandler);

  NfsTimeoutCount = 0;

-- 1.7.9.5

Hi Joe and Wolfgang,

any comment on this?

Best regards, Matthias

-- --- motzblog.wordpress.com

Enric Balletbo Serra

12 Apr 12 Apr

10:08 a.m.

New subject: [U-Boot] [PATCH] net: nfs: add dynamic wait period

Hi all,

2013/1/26 Matthias Brugger matthias.bgg@gmail.com

...

2012/12/11 Matthias Brugger matthias.bgg@gmail.com:

...

This patch tackles the time out problem which leads to break the boot process, when loading file over nfs. The patch does two things.

First of all, we just ignore messages that arrive with a rpc_id smaller then the client id. We just interpret this messages as answers to formaly timed out messages.

Second, when a time out occurs we double the time to wait, so that we do not stress the server resending the last message.

Any comment on the patch?

...

Signed-off-by: Matthias Brugger matthias.bgg@gmail.com

net/nfs.c | 73

+++++++++++++++++++++++++++++++++++++++++++------------------

...

1 file changed, 52 insertions(+), 21 deletions(-)

diff --git a/net/nfs.c b/net/nfs.c index 7f2393f..84aeda1 100644 --- a/net/nfs.c +++ b/net/nfs.c @@ -37,10 +37,14 @@ # define NFS_TIMEOUT CONFIG_NFS_TIMEOUT #endif

+#define NFS_RPC_ERR 1 +#define NFS_RPC_DROP 124

static int fs_mounted; static unsigned long rpc_id; static int nfs_offset = -1; static int nfs_len; +static ulong nfs_timeout = NFS_TIMEOUT;

    debug("%s\n", __func__);

  if (ntohl(rpc_pkt.u.reply.id) != rpc_id)

```
          return -1;
```

  if (ntohl(rpc_pkt.u.reply.id) > rpc_id)

```
          return -NFS_RPC_ERR;
```

  else if (ntohl(rpc_pkt.u.reply.id) < rpc_id)

          return -NFS_RPC_DROP;

  if (rpc_pkt.u.reply.rstatus  ||
      rpc_pkt.u.reply.verifier ||

@@ -428,8 +434,10 @@ nfs_mount_reply(uchar *pkt, unsigned len)

    memcpy((unsigned char *)&rpc_pkt, pkt, len);

  if (ntohl(rpc_pkt.u.reply.id) != rpc_id)

```
          return -1;
```

  if (ntohl(rpc_pkt.u.reply.id) > rpc_id)

```
          return -NFS_RPC_ERR;
```

  else if (ntohl(rpc_pkt.u.reply.id) < rpc_id)

          return -NFS_RPC_DROP;

  if (rpc_pkt.u.reply.rstatus  ||
      rpc_pkt.u.reply.verifier ||

@@ -452,8 +460,10 @@ nfs_umountall_reply(uchar *pkt, unsigned len)

    memcpy((unsigned char *)&rpc_pkt, pkt, len);

  if (ntohl(rpc_pkt.u.reply.id) != rpc_id)

```
          return -1;
```

  if (ntohl(rpc_pkt.u.reply.id) > rpc_id)

```
          return -NFS_RPC_ERR;
```

  else if (ntohl(rpc_pkt.u.reply.id) < rpc_id)

          return -NFS_RPC_DROP;

  if (rpc_pkt.u.reply.rstatus  ||
      rpc_pkt.u.reply.verifier ||

@@ -475,8 +485,10 @@ nfs_lookup_reply(uchar *pkt, unsigned len)

    memcpy((unsigned char *)&rpc_pkt, pkt, len);

  if (ntohl(rpc_pkt.u.reply.id) != rpc_id)

```
          return -1;
```

  if (ntohl(rpc_pkt.u.reply.id) > rpc_id)

```
          return -NFS_RPC_ERR;
```

  else if (ntohl(rpc_pkt.u.reply.id) < rpc_id)

          return -NFS_RPC_DROP;

  if (rpc_pkt.u.reply.rstatus  ||
      rpc_pkt.u.reply.verifier ||

@@ -499,8 +511,10 @@ nfs_readlink_reply(uchar *pkt, unsigned len)

    memcpy((unsigned char *)&rpc_pkt, pkt, len);

  if (ntohl(rpc_pkt.u.reply.id) != rpc_id)

```
          return -1;
```

  if (ntohl(rpc_pkt.u.reply.id) > rpc_id)

```
          return -NFS_RPC_ERR;
```

  else if (ntohl(rpc_pkt.u.reply.id) < rpc_id)

          return -NFS_RPC_DROP;

  if (rpc_pkt.u.reply.rstatus  ||
      rpc_pkt.u.reply.verifier ||

@@ -534,8 +548,10 @@ nfs_read_reply(uchar *pkt, unsigned len)

    memcpy((uchar *)&rpc_pkt, pkt, sizeof(rpc_pkt.u.reply));

  if (ntohl(rpc_pkt.u.reply.id) != rpc_id)

```
          return -1;
```

  if (ntohl(rpc_pkt.u.reply.id) > rpc_id)

```
          return -NFS_RPC_ERR;
```

  else if (ntohl(rpc_pkt.u.reply.id) < rpc_id)

          return -NFS_RPC_DROP;

  if (rpc_pkt.u.reply.rstatus  ||
      rpc_pkt.u.reply.verifier ||

@@ -574,7 +590,7 @@ NfsTimeout(void) NetStartAgain(); } else { puts("T ");

          NetSetTimeout(NFS_TIMEOUT, NfsTimeout);

          NetSetTimeout(nfs_timeout + NFS_TIMEOUT *

NfsTimeoutCount, NfsTimeout);

...

            NfsSend();
    }
} @@ -583,6 +599,7 @@ static void NfsHandler(uchar *pkt, unsigned dest, IPaddr_t sip, unsigned src,

unsigned len)

...

{ int rlen;
  int reply;

  debug("%s\n", __func__);
@@ -591,19 +608,24 @@ NfsHandler(uchar *pkt, unsigned dest, IPaddr_t

sip, unsigned src, unsigned len)

...

    switch (NfsState) {
    case STATE_PRCLOOKUP_PROG_MOUNT_REQ:

          rpc_lookup_reply(PROG_MOUNT, pkt, len);

          if (rpc_lookup_reply(PROG_MOUNT, pkt, len) ==

-NFS_RPC_DROP)

...

                  break;
          NfsState = STATE_PRCLOOKUP_PROG_NFS_REQ;
          NfsSend();
          break;

  case STATE_PRCLOOKUP_PROG_NFS_REQ:

          rpc_lookup_reply(PROG_NFS, pkt, len);

          if (rpc_lookup_reply(PROG_NFS, pkt, len) ==

-NFS_RPC_DROP)

...

                  break;
          NfsState = STATE_MOUNT_REQ;
          NfsSend();
          break;

  case STATE_MOUNT_REQ:

          if (nfs_mount_reply(pkt, len)) {

          reply = nfs_mount_reply(pkt, len);

```
          if (reply == -NFS_RPC_DROP)
```
```
                  break;
```

          else if (reply == -NFS_RPC_ERR) {
                  puts("*** ERROR: Cannot mount\n");
                  /* just to be sure... */
                  NfsState = STATE_UMOUNT_REQ;

@@ -615,7 +637,10 @@ NfsHandler(uchar *pkt, unsigned dest, IPaddr_t sip,

unsigned src, unsigned len)

...

            break;

    case STATE_UMOUNT_REQ:

          if (nfs_umountall_reply(pkt, len)) {

          reply = nfs_umountall_reply(pkt, len);

```
          if (reply == -NFS_RPC_DROP)
```
```
                  break;
```

          else if (reply == -NFS_RPC_ERR) {
                  puts("*** ERROR: Cannot umount\n");
                  net_set_state(NETLOOP_FAIL);
          } else {

@@ -625,7 +650,10 @@ NfsHandler(uchar *pkt, unsigned dest, IPaddr_t sip,

unsigned src, unsigned len)

...

            break;

    case STATE_LOOKUP_REQ:

          if (nfs_lookup_reply(pkt, len)) {

          reply = nfs_lookup_reply(pkt, len);

```
          if (reply == -NFS_RPC_DROP)
```
```
                  break;
```

          else if (reply == -NFS_RPC_ERR) {
                  puts("*** ERROR: File lookup fail\n");
                  NfsState = STATE_UMOUNT_REQ;
                  NfsSend();

@@ -638,7 +666,10 @@ NfsHandler(uchar *pkt, unsigned dest, IPaddr_t sip,

unsigned src, unsigned len)

...

            break;

    case STATE_READLINK_REQ:

          if (nfs_readlink_reply(pkt, len)) {

          reply = nfs_readlink_reply(pkt, len);

```
          if (reply == -NFS_RPC_DROP)
```
```
                  break;
```

          else if (reply == -NFS_RPC_ERR) {
                  puts("*** ERROR: Symlink fail\n");
                  NfsState = STATE_UMOUNT_REQ;
                  NfsSend();

@@ -654,7 +685,7 @@ NfsHandler(uchar *pkt, unsigned dest, IPaddr_t sip,

unsigned src, unsigned len)

...

    case STATE_READ_REQ:
            rlen = nfs_read_reply(pkt, len);

          NetSetTimeout(NFS_TIMEOUT, NfsTimeout);

          NetSetTimeout(nfs_timeout, NfsTimeout);
          if (rlen > 0) {
                  nfs_offset += rlen;
                  NfsSend();

@@ -738,7 +769,7 @@ NfsStart(void) printf("\nLoad address: 0x%lx\n" "Loading: *\b", load_addr);

  NetSetTimeout(NFS_TIMEOUT, NfsTimeout);

  NetSetTimeout(nfs_timeout, NfsTimeout);
  net_set_udp_handler(NfsHandler);

  NfsTimeoutCount = 0;

-- 1.7.9.5

Hi Joe and Wolfgang,

any comment on this?

I had a problem when I tried to load via NFS. Firstly I tried to load the dtb image via NFS, secondly I tried to load the kernel image but if it doesn't work and the file is never loaded.

Applying this patch solved the problem, but seems nobody acked or answered Matthias. Is this patch the proper fix ? In that case it's possible to apply ? If not, any comments how to solve this problem.

Best regards, Enric

...

Best regards, Matthias

--

motzblog.wordpress.com _______________________________________________ U-Boot mailing list U-Boot@lists.denx.de http://lists.denx.de/mailman/listinfo/u-boot

Joe Hershberger

3:56 p.m.

New subject: [U-Boot] [PATCH] net: nfs: add dynamic wait period

Hi Matthias and Enric,

On Fri, Apr 12, 2013 at 3:08 AM, Enric Balletbo Serra eballetbo@gmail.com wrote:

...

Hi all,

2013/1/26 Matthias Brugger matthias.bgg@gmail.com

...
2012/12/11 Matthias Brugger matthias.bgg@gmail.com:

...
This patch tackles the time out problem which leads to break the boot process, when loading file over nfs. The patch does two things.

First of all, we just ignore messages that arrive with a rpc_id smaller then the client id. We just interpret this messages as answers to formaly timed out messages.

Second, when a time out occurs we double the time to wait, so that we do not stress the server resending the last message.

Any comment on the patch?

Hi Joe and Wolfgang,

any comment on this?

I had a problem when I tried to load via NFS. Firstly I tried to load the dtb image via NFS, secondly I tried to load the kernel image but if it doesn't work and the file is never loaded.

Applying this patch solved the problem, but seems nobody acked or answered Matthias. Is this patch the proper fix ? In that case it's possible to apply ? If not, any comments how to solve this problem.

Best regards, Enric

I apologize for the tardiness of dealing with patches this release cycle. I've been swamped at work. I'll get to these soon.

Enric, would you like to send a Tested-by?

Thanks, -Joe

Enric Balletbo Serra

4:11 p.m.

New subject: [U-Boot] [PATCH] net: nfs: add dynamic wait period

Hi Joe,

Thanks for answer.

2013/4/12 Joe Hershberger joe.hershberger@gmail.com

...

Hi Matthias and Enric,

On Fri, Apr 12, 2013 at 3:08 AM, Enric Balletbo Serra eballetbo@gmail.com wrote:

...
Hi all,

2013/1/26 Matthias Brugger matthias.bgg@gmail.com

...
2012/12/11 Matthias Brugger matthias.bgg@gmail.com:

...
This patch tackles the time out problem which leads to break the boot process, when loading file over nfs. The patch does two things.

First of all, we just ignore messages that arrive with a rpc_id

smaller

...
...
...
then the client id. We just interpret this messages as answers to formaly timed out messages.

Second, when a time out occurs we double the time to wait, so that we do not stress the server resending the last message.

Any comment on the patch?

Hi Joe and Wolfgang,

any comment on this?

I had a problem when I tried to load via NFS. Firstly I tried to load the dtb image via NFS, secondly I tried to load the kernel image but if it doesn't work and the file is never loaded.

Applying this patch solved the problem, but seems nobody acked or

answered

...
Matthias. Is this patch the proper fix ? In that case it's possible to

apply

...
? If not, any comments how to solve this problem.

Best regards, Enric

I apologize for the tardiness of dealing with patches this release cycle. I've been swamped at work. I'll get to these soon.

Enric, would you like to send a Tested-by?

Thanks, -Joe

I tested this patch with an IGEPv2 board and an IGEP COM AQUILA.

Tested-by: Enric Balletbo i Serra eballetbo@gmail.com

Joe Hershberger

8 Jul 8 Jul

6:01 p.m.

New subject: [U-Boot] [PATCH] net: nfs: add dynamic wait period

On Tue, Dec 11, 2012 at 12:14 PM, Matthias Brugger matthias.bgg@gmail.com wrote:

...

This patch tackles the time out problem which leads to break the boot process, when loading file over nfs. The patch does two things.

First of all, we just ignore messages that arrive with a rpc_id smaller then the client id. We just interpret this messages as answers to formaly timed out messages.

Second, when a time out occurs we double the time to wait, so that we do not stress the server resending the last message.

Signed-off-by: Matthias Brugger matthias.bgg@gmail.com

Applied, Thanks. -Joe

4320

Age (days ago)

4529

Last active (days ago)

List overview

Download

6 comments

3 participants

tags (0)

participants (3)

Enric Balletbo Serra
Joe Hershberger
Matthias Brugger