[PATCH net] net/smc: fix unexpected SMC_CLC_DECL_ERR_REGRMB error

Subsystems: networking [general], shared memory communications (smc) sockets, the rest

STALE1680d

2 messages, 1 author, 2022-02-25 · open the first message on its own page

[PATCH net] net/smc: fix unexpected SMC_CLC_DECL_ERR_REGRMB error

From: "D. Wythe" <alibuda@linux.alibaba.com>
Date: 2022-02-25 08:11:57

From: "D. Wythe" <alibuda@linux.alibaba.com>

Remove connections from link group is not synchronous with handling
SMC_LLC_DELETE_RKEY, which means that even the number of connections is
less that SMC_RMBS_PER_LGR_MAX, it does not mean that the connection can
register rtoken successfully later, in other words, the rtoken entry may
have not been released. This will cause an unexpected
SMC_CLC_DECL_ERR_REGRMB to be reported, and then ths smc connection have
to fallback to TCP.

Therefore, we need to judge according to the number of idle rtoken
entry.

Fixes: cd6851f30386 ("smc: remote memory buffers (RMBs)")
Signed-off-by: D. Wythe <alibuda@linux.alibaba.com>
---
 net/smc/smc_core.c | 3 ++-
 1 file changed, 2 insertions(+), 1 deletion(-)
diff --git a/net/smc/smc_core.c b/net/smc/smc_core.c
index 29525d0..24ef0af 100644
--- a/net/smc/smc_core.c
+++ b/net/smc/smc_core.c
@@ -1864,7 +1864,8 @@ int smc_conn_create(struct smc_sock *smc, struct smc_init_info *ini)
 		    (ini->smcd_version == SMC_V2 ||
 		     lgr->vlan_id == ini->vlan_id) &&
 		    (role == SMC_CLNT || ini->is_smcd ||
-		     lgr->conns_num < SMC_RMBS_PER_LGR_MAX)) {
+		     lgr->conns_num < SMC_RMBS_PER_LGR_MAX -
+		     bitmap_weight(lgr->rtokens_used_mask, SMC_RMBS_PER_LGR_MAX))) {
 			/* link group found */
 			ini->first_contact_local = 0;
 			conn->lgr = lgr;
-- 
1.8.3.1

Re: [PATCH net] net/smc: fix unexpected SMC_CLC_DECL_ERR_REGRMB error

From: "D. Wythe" <alibuda@linux.alibaba.com>
Date: 2022-02-25 16:45:14


在 2022/2/25 下午4:11, D. Wythe 写道:
quoted hunk
From: "D. Wythe" <alibuda@linux.alibaba.com>

Remove connections from link group is not synchronous with handling
SMC_LLC_DELETE_RKEY, which means that even the number of connections is
less that SMC_RMBS_PER_LGR_MAX, it does not mean that the connection can
register rtoken successfully later, in other words, the rtoken entry may
have not been released. This will cause an unexpected
SMC_CLC_DECL_ERR_REGRMB to be reported, and then ths smc connection have
to fallback to TCP.

Therefore, we need to judge according to the number of idle rtoken
entry.

Fixes: cd6851f30386 ("smc: remote memory buffers (RMBs)")
Signed-off-by: D. Wythe <alibuda@linux.alibaba.com>
---
  net/smc/smc_core.c | 3 ++-
  1 file changed, 2 insertions(+), 1 deletion(-)
diff --git a/net/smc/smc_core.c b/net/smc/smc_core.c
index 29525d0..24ef0af 100644
--- a/net/smc/smc_core.c
+++ b/net/smc/smc_core.c
@@ -1864,7 +1864,8 @@ int smc_conn_create(struct smc_sock *smc, struct smc_init_info *ini)
  		    (ini->smcd_version == SMC_V2 ||
  		     lgr->vlan_id == ini->vlan_id) &&
  		    (role == SMC_CLNT || ini->is_smcd ||
-		     lgr->conns_num < SMC_RMBS_PER_LGR_MAX)) {
+		     lgr->conns_num < SMC_RMBS_PER_LGR_MAX -
+		     bitmap_weight(lgr->rtokens_used_mask, SMC_RMBS_PER_LGR_MAX))) {
  			/* link group found */
  			ini->first_contact_local = 0;
  			conn->lgr = lgr;

I did a horrible math here, i'll send another fix later.

Best wishes.
Keyboard shortcuts
hback out one level
jnext message in thread
kprevious message in thread
ldrill in
Escclose help / fold thread tree
?toggle this help