From: Matthew Wilcox <willy@infradead.org> To: linux-kernel@vger.kernel.org Cc: Matthew Wilcox <willy@infradead.org>, "Nicholas A. Bellinger" <nab@linux-iscsi.org>, Bart Van Assche <bart.vanassche@wdc.com>, Hannes Reinecke <hare@suse.com>, Kees Cook <keescook@chromium.org>, Varun Prakash <varun@chelsio.com>, Sagi Grimberg <sagi@grimberg.me>, Philippe Ombredanne <pombredanne@nexb.com>, Greg Kroah-Hartman <gregkh@linuxfoundation.org>, Kate Stewart <kstewart@linuxfoundation.org>, Thomas Gleixner <tglx@linutronix.de>, "David S. Miller" <davem@davemloft.net>, Denys Vlasenko <dvlasenk@redhat.com>, linux-scsi@vger.kernel.org, target-devel@vger.kernel.org Subject: [PATCH 18/26] target/iscsi: Allocate session IDs from an IDA Date: Thu, 21 Jun 2018 14:28:27 -0700 [thread overview] Message-ID: <20180621212835.5636-19-willy@infradead.org> (raw) In-Reply-To: <20180621212835.5636-1-willy@infradead.org> Since the session is never looked up by ID, we can use the more space-efficient IDA instead of the IDR. I think I found a bug where we never reuse session ID 0, but there may be a good reason for it, so I've merely documented it in this patch. Not reusing session ID 0 doesn't even leak memory. Signed-off-by: Matthew Wilcox <willy@infradead.org> --- drivers/target/iscsi/iscsi_target.c | 10 ++-------- drivers/target/iscsi/iscsi_target.h | 4 +--- drivers/target/iscsi/iscsi_target_login.c | 20 ++++++-------------- 3 files changed, 9 insertions(+), 25 deletions(-) diff --git a/drivers/target/iscsi/iscsi_target.c b/drivers/target/iscsi/iscsi_target.c index 8e223799347a..94bad43c41ff 100644 --- a/drivers/target/iscsi/iscsi_target.c +++ b/drivers/target/iscsi/iscsi_target.c @@ -57,9 +57,8 @@ static DEFINE_SPINLOCK(tiqn_lock); static DEFINE_MUTEX(np_lock); static struct idr tiqn_idr; -struct idr sess_idr; +DEFINE_IDA(sess_ida); struct mutex auth_id_lock; -spinlock_t sess_idr_lock; struct iscsit_global *iscsit_global; @@ -700,9 +699,7 @@ static int __init iscsi_target_init_module(void) spin_lock_init(&iscsit_global->ts_bitmap_lock); mutex_init(&auth_id_lock); - spin_lock_init(&sess_idr_lock); idr_init(&tiqn_idr); - idr_init(&sess_idr); ret = target_register_template(&iscsi_ops); if (ret) @@ -4375,10 +4372,7 @@ int iscsit_close_session(struct iscsi_session *sess) pr_debug("Decremented number of active iSCSI Sessions on" " iSCSI TPG: %hu to %u\n", tpg->tpgt, tpg->nsessions); - spin_lock(&sess_idr_lock); - idr_remove(&sess_idr, sess->session_index); - spin_unlock(&sess_idr_lock); - + ida_free(&sess_ida, sess->session_index); kfree(sess->sess_ops); sess->sess_ops = NULL; spin_unlock_bh(&se_tpg->session_lock); diff --git a/drivers/target/iscsi/iscsi_target.h b/drivers/target/iscsi/iscsi_target.h index 42de1843aa40..48bac0acf8c7 100644 --- a/drivers/target/iscsi/iscsi_target.h +++ b/drivers/target/iscsi/iscsi_target.h @@ -55,9 +55,7 @@ extern struct kmem_cache *lio_ooo_cache; extern struct kmem_cache *lio_qr_cache; extern struct kmem_cache *lio_r2t_cache; -extern struct idr sess_idr; +extern struct ida sess_ida; extern struct mutex auth_id_lock; -extern spinlock_t sess_idr_lock; - #endif /*** ISCSI_TARGET_H ***/ diff --git a/drivers/target/iscsi/iscsi_target_login.c b/drivers/target/iscsi/iscsi_target_login.c index 99501785cdc1..561d2ad38989 100644 --- a/drivers/target/iscsi/iscsi_target_login.c +++ b/drivers/target/iscsi/iscsi_target_login.c @@ -336,22 +336,16 @@ static int iscsi_login_zero_tsih_s1( timer_setup(&sess->time2retain_timer, iscsit_handle_time2retain_timeout, 0); - idr_preload(GFP_KERNEL); - spin_lock_bh(&sess_idr_lock); - ret = idr_alloc(&sess_idr, NULL, 0, 0, GFP_NOWAIT); - if (ret >= 0) - sess->session_index = ret; - spin_unlock_bh(&sess_idr_lock); - idr_preload_end(); - + ret = ida_alloc(&sess_ida, GFP_KERNEL); if (ret < 0) { - pr_err("idr_alloc() for sess_idr failed\n"); + pr_err("Session ID allocation failed %d\n", ret); iscsit_tx_login_rsp(conn, ISCSI_STATUS_CLS_TARGET_ERR, ISCSI_LOGIN_STATUS_NO_RESOURCES); kfree(sess); return -ENOMEM; } + sess->session_index = ret; sess->creation_time = get_jiffies_64(); /* * The FFP CmdSN window values will be allocated from the TPG's @@ -1163,11 +1157,9 @@ void iscsi_target_login_sess_out(struct iscsi_conn *conn, goto old_sess_out; if (conn->sess->se_sess) transport_free_session(conn->sess->se_sess); - if (conn->sess->session_index != 0) { - spin_lock_bh(&sess_idr_lock); - idr_remove(&sess_idr, conn->sess->session_index); - spin_unlock_bh(&sess_idr_lock); - } + /* Um, 0 is a valid ID. I suppose we never free it? */ + if (conn->sess->session_index != 0) + ida_free(&sess_ida, conn->sess->session_index); kfree(conn->sess->sess_ops); kfree(conn->sess); conn->sess = NULL; -- 2.17.1
WARNING: multiple messages have this Message-ID
From: Matthew Wilcox <willy@infradead.org> To: linux-kernel@vger.kernel.org Cc: Matthew Wilcox <willy@infradead.org>, "Nicholas A. Bellinger" <nab@linux-iscsi.org>, Bart Van Assche <bart.vanassche@wdc.com>, Hannes Reinecke <hare@suse.com>, Kees Cook <keescook@chromium.org>, Varun Prakash <varun@chelsio.com>, Sagi Grimberg <sagi@grimberg.me>, Philippe Ombredanne <pombredanne@nexb.com>, Greg Kroah-Hartman <gregkh@linuxfoundation.org>, Kate Stewart <kstewart@linuxfoundation.org>, Thomas Gleixner <tglx@linutronix.de>, "David S. Miller" <davem@davemloft.net>, Denys Vlasenko <dvlasenk@redhat.com>, linux-scsi@vger.kernel.org, target-devel@vger.kernel.org Subject: [PATCH 18/26] target/iscsi: Allocate session IDs from an IDA Date: Thu, 21 Jun 2018 21:28:27 +0000 [thread overview] Message-ID: <20180621212835.5636-19-willy@infradead.org> (raw) In-Reply-To: <20180621212835.5636-1-willy@infradead.org> Since the session is never looked up by ID, we can use the more space-efficient IDA instead of the IDR. I think I found a bug where we never reuse session ID 0, but there may be a good reason for it, so I've merely documented it in this patch. Not reusing session ID 0 doesn't even leak memory. Signed-off-by: Matthew Wilcox <willy@infradead.org> --- drivers/target/iscsi/iscsi_target.c | 10 ++-------- drivers/target/iscsi/iscsi_target.h | 4 +--- drivers/target/iscsi/iscsi_target_login.c | 20 ++++++-------------- 3 files changed, 9 insertions(+), 25 deletions(-) diff --git a/drivers/target/iscsi/iscsi_target.c b/drivers/target/iscsi/iscsi_target.c index 8e223799347a..94bad43c41ff 100644 --- a/drivers/target/iscsi/iscsi_target.c +++ b/drivers/target/iscsi/iscsi_target.c @@ -57,9 +57,8 @@ static DEFINE_SPINLOCK(tiqn_lock); static DEFINE_MUTEX(np_lock); static struct idr tiqn_idr; -struct idr sess_idr; +DEFINE_IDA(sess_ida); struct mutex auth_id_lock; -spinlock_t sess_idr_lock; struct iscsit_global *iscsit_global; @@ -700,9 +699,7 @@ static int __init iscsi_target_init_module(void) spin_lock_init(&iscsit_global->ts_bitmap_lock); mutex_init(&auth_id_lock); - spin_lock_init(&sess_idr_lock); idr_init(&tiqn_idr); - idr_init(&sess_idr); ret = target_register_template(&iscsi_ops); if (ret) @@ -4375,10 +4372,7 @@ int iscsit_close_session(struct iscsi_session *sess) pr_debug("Decremented number of active iSCSI Sessions on" " iSCSI TPG: %hu to %u\n", tpg->tpgt, tpg->nsessions); - spin_lock(&sess_idr_lock); - idr_remove(&sess_idr, sess->session_index); - spin_unlock(&sess_idr_lock); - + ida_free(&sess_ida, sess->session_index); kfree(sess->sess_ops); sess->sess_ops = NULL; spin_unlock_bh(&se_tpg->session_lock); diff --git a/drivers/target/iscsi/iscsi_target.h b/drivers/target/iscsi/iscsi_target.h index 42de1843aa40..48bac0acf8c7 100644 --- a/drivers/target/iscsi/iscsi_target.h +++ b/drivers/target/iscsi/iscsi_target.h @@ -55,9 +55,7 @@ extern struct kmem_cache *lio_ooo_cache; extern struct kmem_cache *lio_qr_cache; extern struct kmem_cache *lio_r2t_cache; -extern struct idr sess_idr; +extern struct ida sess_ida; extern struct mutex auth_id_lock; -extern spinlock_t sess_idr_lock; - #endif /*** ISCSI_TARGET_H ***/ diff --git a/drivers/target/iscsi/iscsi_target_login.c b/drivers/target/iscsi/iscsi_target_login.c index 99501785cdc1..561d2ad38989 100644 --- a/drivers/target/iscsi/iscsi_target_login.c +++ b/drivers/target/iscsi/iscsi_target_login.c @@ -336,22 +336,16 @@ static int iscsi_login_zero_tsih_s1( timer_setup(&sess->time2retain_timer, iscsit_handle_time2retain_timeout, 0); - idr_preload(GFP_KERNEL); - spin_lock_bh(&sess_idr_lock); - ret = idr_alloc(&sess_idr, NULL, 0, 0, GFP_NOWAIT); - if (ret >= 0) - sess->session_index = ret; - spin_unlock_bh(&sess_idr_lock); - idr_preload_end(); - + ret = ida_alloc(&sess_ida, GFP_KERNEL); if (ret < 0) { - pr_err("idr_alloc() for sess_idr failed\n"); + pr_err("Session ID allocation failed %d\n", ret); iscsit_tx_login_rsp(conn, ISCSI_STATUS_CLS_TARGET_ERR, ISCSI_LOGIN_STATUS_NO_RESOURCES); kfree(sess); return -ENOMEM; } + sess->session_index = ret; sess->creation_time = get_jiffies_64(); /* * The FFP CmdSN window values will be allocated from the TPG's @@ -1163,11 +1157,9 @@ void iscsi_target_login_sess_out(struct iscsi_conn *conn, goto old_sess_out; if (conn->sess->se_sess) transport_free_session(conn->sess->se_sess); - if (conn->sess->session_index != 0) { - spin_lock_bh(&sess_idr_lock); - idr_remove(&sess_idr, conn->sess->session_index); - spin_unlock_bh(&sess_idr_lock); - } + /* Um, 0 is a valid ID. I suppose we never free it? */ + if (conn->sess->session_index != 0) + ida_free(&sess_ida, conn->sess->session_index); kfree(conn->sess->sess_ops); kfree(conn->sess); conn->sess = NULL; -- 2.17.1
next prev parent reply other threads:[~2018-06-21 21:32 UTC|newest] Thread overview: 68+ messages / expand[flat|nested] mbox.gz Atom feed top 2018-06-21 21:28 [PATCH 00/26] New IDA API Matthew Wilcox 2018-06-21 21:28 ` [PATCH 01/26] radix tree test suite: fix build Matthew Wilcox 2018-06-21 21:28 ` [PATCH 02/26] ida: Lock the IDA in ida_destroy Matthew Wilcox 2018-06-21 21:28 ` [PATCH 03/26] ida: Add new API Matthew Wilcox 2018-06-21 21:28 ` [PATCH 04/26] mtip32xx: Convert to new IDA API Matthew Wilcox 2018-06-25 8:20 ` Johannes Thumshirn 2018-06-25 8:20 ` Johannes Thumshirn 2018-06-21 21:28 ` [PATCH 05/26] fs: Convert unnamed_dev_ida to new API Matthew Wilcox 2018-06-22 19:45 ` Randy Dunlap 2018-06-22 21:12 ` Matthew Wilcox 2018-06-21 21:28 ` [PATCH 06/26] fs: Convert namespace IDAs " Matthew Wilcox 2018-06-21 21:28 ` [PATCH 07/26] devpts: Convert to new IDA API Matthew Wilcox 2018-06-21 21:28 ` [PATCH 08/26] sd: Convert to new IDA interface Matthew Wilcox 2018-06-21 21:28 ` [PATCH 09/26] osd: Convert to new IDA API Matthew Wilcox 2018-06-21 21:28 ` [PATCH 10/26] rsxx: " Matthew Wilcox 2018-06-21 21:28 ` [PATCH 11/26] cb710: " Matthew Wilcox 2018-06-21 21:33 ` Michał Mirosław 2018-06-21 21:28 ` [PATCH 12/26] Convert net_namespace " Matthew Wilcox 2018-06-21 21:28 ` [PATCH 13/26] ppc: Convert mmu context allocation " Matthew Wilcox 2018-06-22 2:15 ` Nicholas Piggin 2018-06-22 4:38 ` Matthew Wilcox 2018-06-22 4:53 ` Nicholas Piggin 2018-06-22 5:47 ` Aneesh Kumar K.V 2018-06-22 5:47 ` Aneesh Kumar K.V 2018-06-21 21:28 ` [PATCH 14/26] media: Convert entity ID " Matthew Wilcox 2018-07-24 11:05 ` Sakari Ailus 2018-07-30 14:55 ` Mauro Carvalho Chehab 2018-07-31 18:16 ` Matthew Wilcox 2018-06-21 21:28 ` [PATCH 15/26] ppc: Convert vas " Matthew Wilcox 2018-07-05 12:17 ` Matthew Wilcox 2018-06-21 21:28 ` [PATCH 17/26] drm/vmwgfx: Convert " Matthew Wilcox 2018-06-21 21:28 ` Matthew Wilcox [this message] 2018-06-21 21:28 ` [PATCH 18/26] target/iscsi: Allocate session IDs from an IDA Matthew Wilcox 2018-07-26 16:48 ` Mike Christie 2018-07-26 16:48 ` Mike Christie 2018-07-26 16:50 ` Mike Christie 2018-07-26 16:50 ` Mike Christie 2018-07-26 17:13 ` Mike Christie 2018-07-26 17:13 ` Mike Christie 2018-07-26 17:13 ` Mike Christie 2018-07-27 19:38 ` Matthew Wilcox 2018-07-27 19:38 ` Matthew Wilcox 2018-07-27 21:05 ` Mike Christie 2018-07-27 21:05 ` Mike Christie 2018-07-31 2:03 ` Martin K. Petersen 2018-07-31 2:03 ` Martin K. Petersen 2018-07-31 18:15 ` Matthew Wilcox 2018-07-31 18:15 ` Matthew Wilcox 2018-07-31 18:55 ` Mike Christie 2018-07-31 18:55 ` Mike Christie 2018-06-21 21:28 ` [PATCH 19/26] ida: Start new test_ida module Matthew Wilcox 2018-06-21 21:28 ` [PATCH 20/26] idr-test: Convert ida_check_nomem to new API Matthew Wilcox 2018-06-21 21:28 ` [PATCH 21/26] test_ida: Move ida_check_leaf Matthew Wilcox 2018-06-21 21:28 ` [PATCH 22/26] test_ida: Move ida_check_max Matthew Wilcox 2018-06-21 21:28 ` [PATCH 23/26] test_ida: Convert check_ida_conv to new API Matthew Wilcox 2018-06-21 21:28 ` [PATCH 24/26] test_ida: check_ida_destroy and check_ida_alloc Matthew Wilcox 2018-06-21 21:28 ` [PATCH 25/26] ida: Remove old API Matthew Wilcox 2018-06-21 21:28 ` [PATCH 26/26] ida: Change ida_get_new_above to return the id Matthew Wilcox 2018-06-21 21:28 [16/26] dmaengine: Convert to new IDA API Matthew Wilcox 2018-06-21 21:28 ` [PATCH 16/26] " Matthew Wilcox 2018-06-23 12:30 [16/26] " Vinod Koul 2018-06-23 12:30 ` [PATCH 16/26] " Vinod 2018-06-23 23:06 [16/26] " Matthew Wilcox 2018-06-23 23:06 ` [PATCH 16/26] " Matthew Wilcox 2018-06-24 7:57 [16/26] " Lars-Peter Clausen 2018-06-24 7:57 ` [PATCH 16/26] " Lars-Peter Clausen 2018-06-26 23:00 [16/26] " Matthew Wilcox 2018-06-26 23:00 ` [PATCH 16/26] " Matthew Wilcox
Reply instructions: You may reply publicly to this message via plain-text email using any one of the following methods: * Save the following mbox file, import it into your mail client, and reply-to-all from there: mbox Avoid top-posting and favor interleaved quoting: https://en.wikipedia.org/wiki/Posting_style#Interleaved_style * Reply using the --to, --cc, and --in-reply-to switches of git-send-email(1): git send-email \ --in-reply-to=20180621212835.5636-19-willy@infradead.org \ --to=willy@infradead.org \ --cc=bart.vanassche@wdc.com \ --cc=davem@davemloft.net \ --cc=dvlasenk@redhat.com \ --cc=gregkh@linuxfoundation.org \ --cc=hare@suse.com \ --cc=keescook@chromium.org \ --cc=kstewart@linuxfoundation.org \ --cc=linux-kernel@vger.kernel.org \ --cc=linux-scsi@vger.kernel.org \ --cc=nab@linux-iscsi.org \ --cc=pombredanne@nexb.com \ --cc=sagi@grimberg.me \ --cc=target-devel@vger.kernel.org \ --cc=tglx@linutronix.de \ --cc=varun@chelsio.com \ --subject='Re: [PATCH 18/26] target/iscsi: Allocate session IDs from an IDA' \ /path/to/YOUR_REPLY https://kernel.org/pub/software/scm/git/docs/git-send-email.html * If your mail client supports setting the In-Reply-To header via mailto: links, try the mailto: link
This is an external index of several public inboxes, see mirroring instructions on how to clone and mirror all data and code used by this external index.