net/smc: updates 2020-10-07
cover net/smc: updates 2020-10-07
Please apply the following patch series for smc to netdev's net-next tree.
Patch 1 and 2 address warnings from static code checkers, and patch 3 handles a case when all proposed ISM V2 devices fail to init and no V1 devices are tried afterwards.
Karsten Graul (3):
net/smc: consolidate unlocking in same function
net/smc: cleanup buffer usage in smc_listen_work() net/smc: restore smcd_version when all ISM V2 devices failed to init
net/smc/af_smc.c | 92 +++++++++++++++++++++++++----------------------- 1 file changed, 48 insertions(+), 44 deletions(-)
Signature · 1 line
2.17.1
PATCH 1/3 net/smc: consolidate unlocking in same function
Static code checkers warn of inconsistent returns because the lgr mutex
is locked in one function and unlocked in a function called by the
locking function:net/smc/af_smc.c:823 smc_connect_rdma() warn: inconsistent returns 'smc_client_lgr_pending'. net/smc/af_smc.c:897 smc_connect_ism() warn: inconsistent returns 'smc_server_lgr_pending'.
Make the code consistent by doing the unlock in the same function that fetches the lock. No functional changes.
--- net/smc/af_smc.c | 77 +++++++++++++++++++++++++----------------------- 1 file changed, 40 insertions(+), 37 deletions(-)
net/smc/af_smc.c +40 -37
@@ -553,23 +553,12 @@ static int smc_connect_decline_fallback(struct smc_sock *smc, int reason_code, }/* abort connecting */ -static int smc_connect_abort (struct smc_sock *smc,int reason_code, -int local_first) +static void smc_connect_abort (struct smc_sock *smc,int local_first) { -bool is_smcd = smc->conn.lgr->is_smcd; -if (local_first)smc_lgr_cleanup_early (&smc->conn);else smc_conn_free (&smc->conn); -if (is_smcd) -/* there is only one lgr role for SMC-D; use server lock */ -mutex_unlock (&smc_server_lgr_pending); -else -mutex_unlock (&smc_client_lgr_pending); - - smc->connect_nonblock =0 ; -return reason_code; }/* check if there is a rdma device available for this connection. */
@@ -764,43 +753,47 @@ static int smc_connect_rdma(struct smc_sock *smc,break ; } } -if (!link) -return smc_connect_abort (smc,SMC_CLC_DECL_NOSRVLINK , - ini->first_contact_local); +if (!link) { + reason_code =SMC_CLC_DECL_NOSRVLINK ; + goto connect_abort; + } smc->conn.lnk = link; }/* create send buffer and rmb */ -if (smc_buf_create (smc, false)) -return smc_connect_abort (smc,SMC_CLC_DECL_MEM , - ini->first_contact_local); +if (smc_buf_create (smc, false)) { + reason_code =SMC_CLC_DECL_MEM ; + goto connect_abort; + }if (ini->first_contact_local)smc_link_save_peer_info (link, aclc); -if (smc_rmb_rtoken_handling (&smc->conn, link, aclc)) -return smc_connect_abort (smc,SMC_CLC_DECL_ERR_RTOK , - ini->first_contact_local); +if (smc_rmb_rtoken_handling (&smc->conn, link, aclc)) { + reason_code =SMC_CLC_DECL_ERR_RTOK ; + goto connect_abort; + }smc_close_init (smc);smc_rx_init (smc);if (ini->first_contact_local) { -if (smc_ib_ready_link (link)) -return smc_connect_abort (smc,SMC_CLC_DECL_ERR_RDYLNK , - ini->first_contact_local); +if (smc_ib_ready_link (link)) { + reason_code =SMC_CLC_DECL_ERR_RDYLNK ; + goto connect_abort; + } }else { -if (smcr_lgr_reg_rmbs (link, smc->conn.rmb_desc)) -return smc_connect_abort (smc,SMC_CLC_DECL_ERR_REGRMB , - ini->first_contact_local); +if (smcr_lgr_reg_rmbs (link, smc->conn.rmb_desc)) { + reason_code =SMC_CLC_DECL_ERR_REGRMB ; + goto connect_abort; + } }smc_rmb_sync_sg_for_device (&smc->conn); reason_code =smc_clc_send_confirm (smc, ini->first_contact_local,SMC_V1 );if (reason_code) -return smc_connect_abort (smc, reason_code, - ini->first_contact_local); + goto connect_abort;smc_tx_init (smc);
@@ -810,8 +803,7 @@ static int smc_connect_rdma(struct smc_sock *smc, reason_code =smcr_clnt_conf_first_link (smc);smc_llc_flow_stop (link->lgr, &link->lgr->llc_flow_lcl);if (reason_code) -return smc_connect_abort (smc, reason_code, - ini->first_contact_local); + goto connect_abort; }mutex_unlock (&smc_client_lgr_pending);
@@ -821,6 +813,12 @@ static int smc_connect_rdma(struct smc_sock *smc, smc->sk.sk_state =SMC_ACTIVE ;return 0 ; +connect_abort: +smc_connect_abort (smc, ini->first_contact_local); +mutex_unlock (&smc_client_lgr_pending); + smc->connect_nonblock =0 ; + +return reason_code; }/* The server has chosen one of the proposed ISM devices for the communication.
@@ -872,11 +870,10 @@ static int smc_connect_ism(struct smc_sock *smc,/* Create send and receive buffers */ rc =smc_buf_create (smc, true); -if (rc) -return smc_connect_abort (smc, (rc == -ENOSPC ) ? -SMC_CLC_DECL_MAX_DMB : -SMC_CLC_DECL_MEM , - ini->first_contact_local); +if (rc) { + rc = (rc == -ENOSPC ) ?SMC_CLC_DECL_MAX_DMB :SMC_CLC_DECL_MEM ; + goto connect_abort; + }smc_conn_save_peer_info (smc, aclc);smc_close_init (smc);
@@ -886,7 +883,7 @@ static int smc_connect_ism(struct smc_sock *smc, rc =smc_clc_send_confirm (smc, ini->first_contact_local, aclc->hdr.version);if (rc) -return smc_connect_abort (smc, rc, ini->first_contact_local); + goto connect_abort;mutex_unlock (&smc_server_lgr_pending);smc_copy_sock_settings_to_clc (smc);
@@ -895,6 +892,12 @@ static int smc_connect_ism(struct smc_sock *smc, smc->sk.sk_state =SMC_ACTIVE ;return 0 ; +connect_abort: +smc_connect_abort (smc, ini->first_contact_local); +mutex_unlock (&smc_server_lgr_pending); + smc->connect_nonblock =0 ; + +return rc; }/* check if received accept type and version matches a proposed one */
Signature · 1 line
2.17.1
PATCH 2/3 net/smc: cleanup buffer usage in smc_listen_work()
In reply to Karsten Graulcoccinelle informs about
net/smc/af_smc.c:1770:10-11: WARNING: opportunity for kzfree/kvfree_sensitive
Its not that kzfree() would help here, the memset() is done to prepare the buffer for another socket receive.
Fix that warning message by reordering the calls, while at it eliminate the unneeded variable cclc2 and use sizeof(*buf) as above in the same function. No functional changes.
--- net/smc/af_smc.c | 9 +++------ 1 file changed, 3 insertions(+), 6 deletions(-)
net/smc/af_smc.c +3 -6
@@ -1664,7 +1664,6 @@ static void smc_listen_work(struct work_struct *work) smc_listen_work);u8 version = smc_ism_v2_capable ?SMC_V2 :SMC_V1 ;struct socket *newclcsock = new_smc->clcsock; -struct smc_clc_msg_accept_confirm_v2 *cclc2;struct smc_clc_msg_accept_confirm *cclc;struct smc_clc_msg_proposal_area *buf;struct smc_clc_msg_proposal *pclc;
@@ -1740,11 +1739,9 @@ static void smc_listen_work(struct work_struct *work)mutex_unlock (&smc_server_lgr_pending);/* receive SMC Confirm CLC message */ - cclc2 = (struct smc_clc_msg_accept_confirm_v2 *)buf; - cclc = (struct smc_clc_msg_accept_confirm *)cclc2; -memset (buf,0 ,sizeof (struct smc_clc_msg_proposal_area )); - rc =smc_clc_wait_msg (new_smc, cclc2, -sizeof (struct smc_clc_msg_proposal_area ), +memset (buf,0 ,sizeof (*buf)); + cclc = (struct smc_clc_msg_accept_confirm *)buf; + rc =smc_clc_wait_msg (new_smc, cclc,sizeof (*buf),SMC_CLC_CONFIRM ,CLC_WAIT_TIME );if (rc) {if (!ini->is_smcd)
Signature · 1 line
2.17.1
PATCH 3/3 net/smc: restore smcd_version when all ISM V2 devices failed to init
In reply to Karsten GraulField ini->smcd_version is set to SMC_V2 before calling smc_listen_ism_init(). This clears the V1 bit that may be set. When all matching ISM V2 devices fail to initialize then the smcd_version field needs to get restored to allow any possible V1 devices to initialize. And be consistent, always go to the not_found label when no device was found.
--- net/smc/af_smc.c | 6 +++++- 1 file changed, 5 insertions(+), 1 deletion(-)
net/smc/af_smc.c +5 -1
@@ -1481,11 +1481,12 @@ static void smc_find_ism_v2_device_serv(struct smc_sock *new_smc,struct smc_clc_v2_extension *smc_v2_ext;struct smc_clc_msg_smcd *pclc_smcd;unsigned int matches =0 ; +u8 smcd_version;u8 *eid =NULL ;int i;if (!(ini->smcd_version &SMC_V2 ) || !smcd_indicated (ini->smc_type_v2)) -return ; + goto not_found; pclc_smcd =smc_get_clc_msg_smcd (pclc); smc_v2_ext =smc_get_clc_v2_ext (pclc);
@@ -1519,6 +1520,7 @@ static void smc_find_ism_v2_device_serv(struct smc_sock *new_smc, }/* separate - outside the smcd_dev_list.lock */ + smcd_version = ini->smcd_version;for (i =0 ; i < matches; i++) { ini->smcd_version =SMC_V2 ; ini->is_smcd = true;
@@ -1528,6 +1530,8 @@ static void smc_find_ism_v2_device_serv(struct smc_sock *new_smc,continue ;return ;/* matching and usable V2 ISM device found */ } +/* no V2 ISM device could be initialized */ + ini->smcd_version = smcd_version;/* restore original value */ not_found: ini->smcd_version &= ~SMC_V2 ;
Signature · 1 line
2.17.1
Quoted 5 lines · On Wed, 7 Oct 2020 22:57:40 +0200 Karsten Graul wrote:
Please apply the following patch series for smc to netdev's net-next tree.
Patch 1 and 2 address warnings from static code checkers, and patch 3 handles a case when all proposed ISM V2 devices fail to init and no V1 devices are tried afterwards.
Applied, thanks!
I'm inducing the last patch is a fix for code only in net-next. It would still have been better to have a Fixes tag there to make it clear that there is no need to backport.