Skip to contentSkip to conversation

scsi: lpfc: Move NPIV's transport unregistration to after resource clean up

PATCH New Sasha Levin 1 participant 13 messages Apr 7, 2024

Subsystems (8)
lore
Share & tools
lore

Sasha Levin

author maintainer
#0 lore raw

PATCH 1/13 scsi: lpfc: Move NPIV's transport unregistration to after resource clean up

From: Justin Tee <[email protected]>

[ Upstream commit 4ddf01f2f1504fa08b766e8cfeec558e9f8eef6c ]

There are cases after NPIV deletion where the fabric switch still believes the NPIV is logged into the fabric. This occurs when a vport is unregistered before the Remove All DA_ID CT and LOGO ELS are sent to the fabric.

Currently fc_remove_host(), which calls dev_loss_tmo for all D_IDs including the fabric D_ID, removes the last ndlp reference and frees the ndlp rport object. This sometimes causes the race condition where the final DA_ID and LOGO are skipped from being sent to the fabric switch.

Fix by moving the fc_remove_host() and scsi_remove_host() calls after DA_ID and LOGO are sent.

---
 drivers/scsi/lpfc/lpfc_vport.c | 8 ++++----
 1 file changed, 4 insertions(+), 4 deletions(-)
drivers/scsi/lpfc/lpfc_vport.c +4 -4
index 4d171f5c213f7..6b4259894584f 100644
@@ -693,10 +693,6 @@ lpfc_vport_delete(struct fc_vport *fc_vport)
 	lpfc_free_sysfs_attr(vport);
 	lpfc_debugfs_terminate(vport);
 
-	/* Remove FC host to break driver binding. */
-	fc_remove_host(shost);
-	scsi_remove_host(shost);
-
 	/* Send the DA_ID and Fabric LOGO to cleanup Nameserver entries. */
 	ndlp = lpfc_findnode_did(vport, Fabric_DID);
 	if (!ndlp)
@@ -740,6 +736,10 @@ lpfc_vport_delete(struct fc_vport *fc_vport)
 
 skip_logo:
 
+	/* Remove FC host to break driver binding. */
+	fc_remove_host(shost);
+	scsi_remove_host(shost);
+
 	lpfc_cleanup(vport);
 
 	/* Remove scsi host now.  The nodes are cleaned up. */
Signature · 1 line
2.43.0
author maintainer
#1 lore raw

PATCH 2/13 scsi: lpfc: Update lpfc_ramp_down_queue_handler() logic

From: Justin Tee <[email protected]>

[ Upstream commit bb011631435c705cdeddca68d5c85fd40a4320f9 ]

Typically when an out of resource CQE status is detected, the lpfc_ramp_down_queue_handler() logic is called to help reduce I/O load by reducing an sdev's queue_depth.

However, the current lpfc_rampdown_queue_depth() logic does not help reduce queue_depth. num_cmd_success is never updated and is always zero, which means new_queue_depth will always be set to sdev->queue_depth. So, new_queue_depth = sdev->queue_depth - new_queue_depth always sets new_queue_depth to zero. And, scsi_change_queue_depth(sdev, 0) is essentially a no-op.

Change the lpfc_ramp_down_queue_handler() logic to set new_queue_depth equal to sdev->queue_depth subtracted from number of times num_rsrc_err was incremented. If num_rsrc_err is >= sdev->queue_depth, then set new_queue_depth equal to 1. Eventually, the frequency of Good_Status frames will signal SCSI upper layer to auto increase the queue_depth back to the driver default of 64 via scsi_handle_queue_ramp_up().

---
 drivers/scsi/lpfc/lpfc.h      |  1 -
 drivers/scsi/lpfc/lpfc_scsi.c | 13 ++++---------
 2 files changed, 4 insertions(+), 10 deletions(-)
drivers/scsi/lpfc/lpfc.h +0 -1
index dc5ac3cc70f6d..6f08fbe103cb9 100644
@@ -1355,7 +1355,6 @@ struct lpfc_hba {
 	struct timer_list fabric_block_timer;
 	unsigned long bit_flags;
 	atomic_t num_rsrc_err;
-	atomic_t num_cmd_success;
 	unsigned long last_rsrc_error_time;
 	unsigned long last_ramp_down_time;
 #ifdef CONFIG_SCSI_LPFC_DEBUG_FS
drivers/scsi/lpfc/lpfc_scsi.c +4 -9
index 0bb7e164b525f..2a81a42de5c14 100644
@@ -167,11 +167,10 @@ lpfc_ramp_down_queue_handler(struct lpfc_hba *phba)
 	struct Scsi_Host  *shost;
 	struct scsi_device *sdev;
 	unsigned long new_queue_depth;
-	unsigned long num_rsrc_err, num_cmd_success;
+	unsigned long num_rsrc_err;
 	int i;
 
 	num_rsrc_err = atomic_read(&phba->num_rsrc_err);
-	num_cmd_success = atomic_read(&phba->num_cmd_success);
 
 	/*
 	 * The error and success command counters are global per
@@ -186,20 +185,16 @@ lpfc_ramp_down_queue_handler(struct lpfc_hba *phba)
 		for (i = 0; i <= phba->max_vports && vports[i] != NULL; i++) {
 			shost = lpfc_shost_from_vport(vports[i]);
 			shost_for_each_device(sdev, shost) {
-				new_queue_depth =
-					sdev->queue_depth * num_rsrc_err /
-					(num_rsrc_err + num_cmd_success);
-				if (!new_queue_depth)
-					new_queue_depth = sdev->queue_depth - 1;
+				if (num_rsrc_err >= sdev->queue_depth)
+					new_queue_depth = 1;
 				else
 					new_queue_depth = sdev->queue_depth -
-								new_queue_depth;
+						num_rsrc_err;
 				scsi_change_queue_depth(sdev, new_queue_depth);
 			}
 		}
 	lpfc_destroy_vport_work_array(phba, vports);
 	atomic_set(&phba->num_rsrc_err, 0);
-	atomic_set(&phba->num_cmd_success, 0);
 }
 
 /**
Signature · 1 line
2.43.0
author maintainer
#2 lore raw

PATCH 3/13 scsi: lpfc: Replace hbalock with ndlp lock in lpfc_nvme_unregister_port()

In reply to Sasha Levin

From: Justin Tee <[email protected]>

[ Upstream commit d11272be497e48a8e8f980470eb6b70e92eed0ce ]

The ndlp object update in lpfc_nvme_unregister_port() should be protected by the ndlp lock rather than hbalock.

---
 drivers/scsi/lpfc/lpfc_nvme.c | 4 ++--
 1 file changed, 2 insertions(+), 2 deletions(-)
drivers/scsi/lpfc/lpfc_nvme.c +2 -2
index 152245f7cacc7..7e9e0d969256a 100644
@@ -2621,9 +2621,9 @@ lpfc_nvme_unregister_port(struct lpfc_vport *vport, struct lpfc_nodelist *ndlp)
 		/* No concern about the role change on the nvme remoteport.
 		 * The transport will update it.
 		 */
-		spin_lock_irq(&vport->phba->hbalock);
+		spin_lock_irq(&ndlp->lock);
 		ndlp->fc4_xpt_flags |= NVME_XPT_UNREG_WAIT;
-		spin_unlock_irq(&vport->phba->hbalock);
+		spin_unlock_irq(&ndlp->lock);
 
 		/* Don't let the host nvme transport keep sending keep-alives
 		 * on this remoteport. Vport is unloading, no recovery. The
Signature · 1 line
2.43.0
author maintainer
#3 lore raw

PATCH 4/13 scsi: lpfc: Release hbalock before calling lpfc_worker_wake_up()

In reply to Sasha Levin

From: Justin Tee <[email protected]>

[ Upstream commit ded20192dff31c91cef2a04f7e20e60e9bb887d3 ]

lpfc_worker_wake_up() calls the lpfc_work_done() routine, which takes the hbalock. Thus, lpfc_worker_wake_up() should not be called while holding the hbalock to avoid potential deadlock.

---
 drivers/scsi/lpfc/lpfc_els.c     | 20 ++++++++++----------
 drivers/scsi/lpfc/lpfc_hbadisc.c |  5 ++---
 drivers/scsi/lpfc/lpfc_sli.c     | 14 +++++++-------
 3 files changed, 19 insertions(+), 20 deletions(-)
drivers/scsi/lpfc/lpfc_els.c +10 -10
index 6b5ce9869e6b4..05764008f6e70 100644
@@ -4384,23 +4384,23 @@ lpfc_els_retry_delay(struct timer_list *t)
 	unsigned long flags;
 	struct lpfc_work_evt  *evtp = &ndlp->els_retry_evt;
 
+	/* Hold a node reference for outstanding queued work */
+	if (!lpfc_nlp_get(ndlp))
+		return;
+
 	spin_lock_irqsave(&phba->hbalock, flags);
 	if (!list_empty(&evtp->evt_listp)) {
 		spin_unlock_irqrestore(&phba->hbalock, flags);
+		lpfc_nlp_put(ndlp);
 		return;
 	}
 
-	/* We need to hold the node by incrementing the reference
-	 * count until the queued work is done
-	 */
-	evtp->evt_arg1  = lpfc_nlp_get(ndlp);
-	if (evtp->evt_arg1) {
-		evtp->evt = LPFC_EVT_ELS_RETRY;
-		list_add_tail(&evtp->evt_listp, &phba->work_list);
-		lpfc_worker_wake_up(phba);
-	}
+	evtp->evt_arg1 = ndlp;
+	evtp->evt = LPFC_EVT_ELS_RETRY;
+	list_add_tail(&evtp->evt_listp, &phba->work_list);
 	spin_unlock_irqrestore(&phba->hbalock, flags);
-	return;
+
+	lpfc_worker_wake_up(phba);
 }
 
 /**
drivers/scsi/lpfc/lpfc_hbadisc.c +2 -3
index 549fa7d6c0f6f..aaa98a006fdcb 100644
@@ -241,7 +241,9 @@ lpfc_dev_loss_tmo_callbk(struct fc_rport *rport)
 		if (evtp->evt_arg1) {
 			evtp->evt = LPFC_EVT_DEV_LOSS;
 			list_add_tail(&evtp->evt_listp, &phba->work_list);
+			spin_unlock_irqrestore(&phba->hbalock, iflags);
 			lpfc_worker_wake_up(phba);
+			return;
 		}
 		spin_unlock_irqrestore(&phba->hbalock, iflags);
 	} else {
@@ -259,10 +261,7 @@ lpfc_dev_loss_tmo_callbk(struct fc_rport *rport)
 			lpfc_disc_state_machine(vport, ndlp, NULL,
 						NLP_EVT_DEVICE_RM);
 		}
-
 	}
-
-	return;
 }
 
 /**
drivers/scsi/lpfc/lpfc_sli.c +7 -7
index 427a6ac803e50..47b8102a7063a 100644
@@ -1217,9 +1217,9 @@ lpfc_set_rrq_active(struct lpfc_hba *phba, struct lpfc_nodelist *ndlp,
 	empty = list_empty(&phba->active_rrq_list);
 	list_add_tail(&rrq->list, &phba->active_rrq_list);
 	phba->hba_flag |= HBA_RRQ_ACTIVE;
+	spin_unlock_irqrestore(&phba->hbalock, iflags);
 	if (empty)
 		lpfc_worker_wake_up(phba);
-	spin_unlock_irqrestore(&phba->hbalock, iflags);
 	return 0;
 out:
 	spin_unlock_irqrestore(&phba->hbalock, iflags);
@@ -11361,18 +11361,18 @@ lpfc_sli_post_recovery_event(struct lpfc_hba *phba,
 	unsigned long iflags;
 	struct lpfc_work_evt  *evtp = &ndlp->recovery_evt;
 
+	/* Hold a node reference for outstanding queued work */
+	if (!lpfc_nlp_get(ndlp))
+		return;
+
 	spin_lock_irqsave(&phba->hbalock, iflags);
 	if (!list_empty(&evtp->evt_listp)) {
 		spin_unlock_irqrestore(&phba->hbalock, iflags);
+		lpfc_nlp_put(ndlp);
 		return;
 	}
 
-	/* Incrementing the reference count until the queued work is done. */
-	evtp->evt_arg1  = lpfc_nlp_get(ndlp);
-	if (!evtp->evt_arg1) {
-		spin_unlock_irqrestore(&phba->hbalock, iflags);
-		return;
-	}
+	evtp->evt_arg1 = ndlp;
 	evtp->evt = LPFC_EVT_RECOVER_PORT;
 	list_add_tail(&evtp->evt_listp, &phba->work_list);
 	spin_unlock_irqrestore(&phba->hbalock, iflags);
Signature · 1 line
2.43.0
author maintainer
#4 lore raw

PATCH 5/13 gfs2: Fix invalid metadata access in punch_hole

In reply to Sasha Levin

From: Andrew Price <[email protected]>

[ Upstream commit c95346ac918c5badf51b9a7ac58a26d3bd5bb224 ]

In punch_hole(), when the offset lies in the final block for a given height, there is no hole to punch, but the maximum size check fails to detect that. Consequently, punch_hole() will try to punch a hole beyond the end of the metadata and fail. Fix the maximum size check.

---
 fs/gfs2/bmap.c | 5 +++--
 1 file changed, 3 insertions(+), 2 deletions(-)
fs/gfs2/bmap.c +3 -2
index e7537fd305dd2..9ad11e5bf14c3 100644
@@ -1702,7 +1702,8 @@ static int punch_hole(struct gfs2_inode *ip, u64 offset, u64 length)
 	struct buffer_head *dibh, *bh;
 	struct gfs2_holder rd_gh;
 	unsigned int bsize_shift = sdp->sd_sb.sb_bsize_shift;
-	u64 lblock = (offset + (1 << bsize_shift) - 1) >> bsize_shift;
+	unsigned int bsize = 1 << bsize_shift;
+	u64 lblock = (offset + bsize - 1) >> bsize_shift;
 	__u16 start_list[GFS2_MAX_META_HEIGHT];
 	__u16 __end_list[GFS2_MAX_META_HEIGHT], *end_list = NULL;
 	unsigned int start_aligned, end_aligned;
@@ -1713,7 +1714,7 @@ static int punch_hole(struct gfs2_inode *ip, u64 offset, u64 length)
 	u64 prev_bnr = 0;
 	__be64 *start, *end;
 
-	if (offset >= maxsize) {
+	if (offset + bsize - 1 >= maxsize) {
 		/*
 		 * The starting point lies beyond the allocated meta-data;
 		 * there are no blocks do deallocate.
Signature · 1 line
2.43.0
author maintainer
#5 lore raw

PATCH 6/13 wifi: mac80211: fix ieee80211_bss_*_flags kernel-doc

In reply to Sasha Levin

From: Jeff Johnson <[email protected]>

[ Upstream commit 774f8841f55d7ac4044c79812691649da203584a ]

Running kernel-doc on ieee80211_i.h flagged the following:
net/mac80211/ieee80211_i.h:145: warning: expecting prototype for enum ieee80211_corrupt_data_flags. Prototype was for enum ieee80211_bss_corrupt_data_flags instead net/mac80211/ieee80211_i.h:162: warning: expecting prototype for enum ieee80211_valid_data_flags. Prototype was for enum ieee80211_bss_valid_data_flags instead

Fix these warnings.

---
 net/mac80211/ieee80211_i.h | 4 ++--
 1 file changed, 2 insertions(+), 2 deletions(-)
net/mac80211/ieee80211_i.h +2 -2
index d5dd2d9e89b48..3e14d5c9aa1b4 100644
@@ -120,7 +120,7 @@ struct ieee80211_bss {
 };
 
 /**
- * enum ieee80211_corrupt_data_flags - BSS data corruption flags
+ * enum ieee80211_bss_corrupt_data_flags - BSS data corruption flags
  * @IEEE80211_BSS_CORRUPT_BEACON: last beacon frame received was corrupted
  * @IEEE80211_BSS_CORRUPT_PROBE_RESP: last probe response received was corrupted
  *
@@ -133,7 +133,7 @@ enum ieee80211_bss_corrupt_data_flags {
 };
 
 /**
- * enum ieee80211_valid_data_flags - BSS valid data flags
+ * enum ieee80211_bss_valid_data_flags - BSS valid data flags
  * @IEEE80211_BSS_VALID_WMM: WMM/UAPSD data was gathered from non-corrupt IE
  * @IEEE80211_BSS_VALID_RATES: Supported rates were gathered from non-corrupt IE
  * @IEEE80211_BSS_VALID_ERP: ERP flag was gathered from non-corrupt IE
Signature · 1 line
2.43.0
author maintainer
#6 lore raw

PATCH 7/13 wifi: cfg80211: fix rdev_dump_mpp() arguments order

In reply to Sasha Levin

From: Igor Artemiev <[email protected]>

[ Upstream commit ec50f3114e55406a1aad24b7dfaa1c3f4336d8eb ]

Fix the order of arguments in the TP_ARGS macro
for the rdev_dump_mpp tracepoint event.

Found by Linux Verification Center (linuxtesting.org).

---
 net/wireless/trace.h | 2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)
net/wireless/trace.h +1 -1
index a405c3edbc47e..cb5c3224e038a 100644
@@ -1018,7 +1018,7 @@ TRACE_EVENT(rdev_get_mpp,
 TRACE_EVENT(rdev_dump_mpp,
 	TP_PROTO(struct wiphy *wiphy, struct net_device *netdev, int _idx,
 		 u8 *dst, u8 *mpp),
-	TP_ARGS(wiphy, netdev, _idx, mpp, dst),
+	TP_ARGS(wiphy, netdev, _idx, dst, mpp),
 	TP_STRUCT__entry(
 		WIPHY_ENTRY
 		NETDEV_ENTRY
Signature · 1 line
2.43.0
author maintainer
#7 lore raw

PATCH 8/13 net: mark racy access on sk->sk_rcvbuf

In reply to Sasha Levin

From: linke li <[email protected]>

[ Upstream commit c2deb2e971f5d9aca941ef13ee05566979e337a4 ]

sk->sk_rcvbuf in __sock_queue_rcv_skb() and __sk_receive_skb() can be changed by other threads. Mark this as benign using READ_ONCE().

This patch is aimed at reducing the number of benign races reported by KCSAN in order to focus future debugging effort on harmful races.

---
 net/core/sock.c | 4 ++--
 1 file changed, 2 insertions(+), 2 deletions(-)
net/core/sock.c +2 -2
index c8803b95ea0da..054fbace08e0a 100644
@@ -481,7 +481,7 @@ int __sock_queue_rcv_skb(struct sock *sk, struct sk_buff *skb)
 	unsigned long flags;
 	struct sk_buff_head *list = &sk->sk_receive_queue;
 
-	if (atomic_read(&sk->sk_rmem_alloc) >= sk->sk_rcvbuf) {
+	if (atomic_read(&sk->sk_rmem_alloc) >= READ_ONCE(sk->sk_rcvbuf)) {
 		atomic_inc(&sk->sk_drops);
 		trace_sock_rcvqueue_full(sk, skb);
 		return -ENOMEM;
@@ -551,7 +551,7 @@ int __sk_receive_skb(struct sock *sk, struct sk_buff *skb,
 
 	skb->dev = NULL;
 
-	if (sk_rcvqueues_full(sk, sk->sk_rcvbuf)) {
+	if (sk_rcvqueues_full(sk, READ_ONCE(sk->sk_rcvbuf))) {
 		atomic_inc(&sk->sk_drops);
 		goto discard_and_relse;
 	}
Signature · 1 line
2.43.0
author maintainer
#8 lore raw

PATCH 9/13 scsi: mpi3mr: Avoid memcpy field-spanning write WARNING

In reply to Sasha Levin

From: Shin'ichiro Kawasaki <[email protected]>

[ Upstream commit 429846b4b6ce9853e0d803a2357bb2e55083adf0 ]

When the "storcli2 show" command is executed for eHBA-9600, mpi3mr driver prints this WARNING message:

memcpy: detected field-spanning write (size 128) of single field "bsg_reply_buf->reply_buf" at drivers/scsi/mpi3mr/mpi3mr_app.c:1658 (size 1)

  WARNING: CPU: 0 PID: 12760 at drivers/scsi/mpi3mr/mpi3mr_app.c:1658 mpi3mr_bsg_request+0x6b12/0x7f10 [mpi3mr]

The cause of the WARN is 128 bytes memcpy to the 1 byte size array "__u8 replay_buf[1]" in the struct mpi3mr_bsg_in_reply_buf. The array is intended to be a flexible length array, so the WARN is a false positive.

To suppress the WARN, remove the constant number '1' from the array declaration and clarify that it has flexible length. Also, adjust the memory allocation size to match the change.

---
 drivers/scsi/mpi3mr/mpi3mr_app.c    | 2 +-
 include/uapi/scsi/scsi_bsg_mpi3mr.h | 2 +-
 2 files changed, 2 insertions(+), 2 deletions(-)
drivers/scsi/mpi3mr/mpi3mr_app.c +1 -1
index 8c662d08706f1..42600e5c457a1 100644
@@ -1344,7 +1344,7 @@ static long mpi3mr_bsg_process_mpt_cmds(struct bsg_job *job, unsigned int *reply
 	if ((mpirep_offset != 0xFF) &&
 	    drv_bufs[mpirep_offset].bsg_buf_len) {
 		drv_buf_iter = &drv_bufs[mpirep_offset];
-		drv_buf_iter->kern_buf_len = (sizeof(*bsg_reply_buf) - 1 +
+		drv_buf_iter->kern_buf_len = (sizeof(*bsg_reply_buf) +
 					   mrioc->reply_sz);
 		bsg_reply_buf = kzalloc(drv_buf_iter->kern_buf_len, GFP_KERNEL);
 
include/uapi/scsi/scsi_bsg_mpi3mr.h +1 -1
index fdc3517f9e199..c48c5d08c0fa0 100644
@@ -382,7 +382,7 @@ struct mpi3mr_bsg_in_reply_buf {
 	__u8	mpi_reply_type;
 	__u8	rsvd1;
 	__u16	rsvd2;
-	__u8	reply_buf[1];
+	__u8	reply_buf[];
 };
 
 /**
Signature · 1 line
2.43.0
author maintainer
#9 lore raw

PATCH 10/13 scsi: bnx2fc: Remove spin_lock_bh while releasing resources after upload

In reply to Sasha Levin

From: Saurav Kashyap <[email protected]>

[ Upstream commit c214ed2a4dda35b308b0b28eed804d7ae66401f9 ]

The session resources are used by FW and driver when session is offloaded, once session is uploaded these resources are not used. The lock is not required as these fields won't be used any longer. The offload and upload calls are sequential, hence lock is not required.

This will suppress following BUG_ON():

[  449.843143] ------------[ cut here ]------------
[  449.848302] kernel BUG at mm/vmalloc.c:2727!
[  449.853072] invalid opcode: 0000 [#1] PREEMPT SMP PTI
[  449.858712] CPU: 5 PID: 1996 Comm: kworker/u24:2 Not tainted 5.14.0-118.el9.x86_64 #1
Rebooting.
[  449.867454] Hardware name: Dell Inc. PowerEdge R730/0WCJNT, BIOS 2.3.4 11/08/2016
[  449.876966] Workqueue: fc_rport_eq fc_rport_work [libfc]
[  449.882910] RIP: 0010:vunmap+0x2e/0x30
[  449.887098] Code: 00 65 8b 05 14 a2 f0 4a a9 00 ff ff 00 75 1b 55 48 89 fd e8 34 36 79 00 48 85 ed 74 0b 48 89 ef 31 f6 5d e9 14 fc ff ff 5d c3 <0f> 0b 0f 1f 44 00 00 41 57 41 56 49 89 ce 41 55 49 89 fd 41 54 41
[  449.908054] RSP: 0018:ffffb83d878b3d68 EFLAGS: 00010206
[  449.913887] RAX: 0000000080000201 RBX: ffff8f4355133550 RCX: 000000000d400005
[  449.921843] RDX: 0000000000000001 RSI: 0000000000001000 RDI: ffffb83da53f5000
[  449.929808] RBP: ffff8f4ac6675800 R08: ffffb83d878b3d30 R09: 00000000000efbdf
[  449.937774] R10: 0000000000000003 R11: ffff8f434573e000 R12: 0000000000001000
[  449.945736] R13: 0000000000001000 R14: ffffb83da53f5000 R15: ffff8f43d4ea3ae0
[  449.953701] FS:  0000000000000000(0000) GS:ffff8f529fc80000(0000) knlGS:0000000000000000
[  449.962732] CS:  0010 DS: 0000 ES: 0000 CR0: 0000000080050033
[  449.969138] CR2: 00007f8cf993e150 CR3: 0000000efbe10003 CR4: 00000000003706e0
[  449.977102] DR0: 0000000000000000 DR1: 0000000000000000 DR2: 0000000000000000
[  449.985065] DR3: 0000000000000000 DR6: 00000000fffe0ff0 DR7: 0000000000000400
[  449.993028] Call Trace:
[  449.995756]  __iommu_dma_free+0x96/0x100
[  450.000139]  bnx2fc_free_session_resc+0x67/0x240 [bnx2fc]
[  450.006171]  bnx2fc_upload_session+0xce/0x100 [bnx2fc]
[  450.011910]  bnx2fc_rport_event_handler+0x9f/0x240 [bnx2fc]
[  450.018136]  fc_rport_work+0x103/0x5b0 [libfc]
[  450.023103]  process_one_work+0x1e8/0x3c0
[  450.027581]  worker_thread+0x50/0x3b0
[  450.031669]  ? rescuer_thread+0x370/0x370
[  450.036143]  kthread+0x149/0x170
[  450.039744]  ? set_kthread_struct+0x40/0x40
[  450.044411]  ret_from_fork+0x22/0x30
[  450.048404] Modules linked in: vfat msdos fat xfs nfs_layout_nfsv41_files rpcsec_gss_krb5 auth_rpcgss nfsv4 dns_resolver dm_service_time qedf qed crc8 bnx2fc libfcoe libfc scsi_transport_fc intel_rapl_msr intel_rapl_common x86_pkg_temp_thermal intel_powerclamp dcdbas rapl intel_cstate intel_uncore mei_me pcspkr mei ipmi_ssif lpc_ich ipmi_si fuse zram ext4 mbcache jbd2 loop nfsv3 nfs_acl nfs lockd grace fscache netfs irdma ice sd_mod t10_pi sg ib_uverbs ib_core 8021q garp mrp stp llc mgag200 i2c_algo_bit drm_kms_helper syscopyarea sysfillrect sysimgblt mxm_wmi fb_sys_fops cec crct10dif_pclmul ahci crc32_pclmul bnx2x drm ghash_clmulni_intel libahci rfkill i40e libata megaraid_sas mdio wmi sunrpc lrw dm_crypt dm_round_robin dm_multipath dm_snapshot dm_bufio dm_mirror dm_region_hash dm_log dm_zero dm_mod linear raid10 raid456 async_raid6_recov async_memcpy async_pq async_xor async_tx raid6_pq libcrc32c crc32c_intel raid1 raid0 iscsi_ibft squashfs be2iscsi bnx2i cnic uio cxgb4i cxgb4 tls
[  450.048497]  libcxgbi libcxgb qla4xxx iscsi_boot_sysfs iscsi_tcp libiscsi_tcp libiscsi scsi_transport_iscsi edd ipmi_devintf ipmi_msghandler
[  450.159753] ---[ end trace 712de2c57c64abc8 ]---
---
 drivers/scsi/bnx2fc/bnx2fc_tgt.c | 2 --
 1 file changed, 2 deletions(-)
drivers/scsi/bnx2fc/bnx2fc_tgt.c +0 -2
index 2c246e80c1c4d..d91659811eb3c 100644
@@ -833,7 +833,6 @@ static void bnx2fc_free_session_resc(struct bnx2fc_hba *hba,
 
 	BNX2FC_TGT_DBG(tgt, "Freeing up session resources\n");
 
-	spin_lock_bh(&tgt->cq_lock);
 	ctx_base_ptr = tgt->ctx_base;
 	tgt->ctx_base = NULL;
 
@@ -889,7 +888,6 @@ static void bnx2fc_free_session_resc(struct bnx2fc_hba *hba,
 				    tgt->sq, tgt->sq_dma);
 		tgt->sq = NULL;
 	}
-	spin_unlock_bh(&tgt->cq_lock);
 
 	if (ctx_base_ptr)
 		iounmap(ctx_base_ptr);
Signature · 1 line
2.43.0
author maintainer
#10 lore raw

PATCH 11/13 btrfs: return accurate error code on open failure in open_fs_devices()

In reply to Sasha Levin

From: Anand Jain <[email protected]>

[ Upstream commit 2f1aeab9fca1a5f583be1add175d1ee95c213cfa ]

When attempting to exclusive open a device which has no exclusive open permission, such as a physical device associated with the flakey dm device, the open operation will fail, resulting in a mount failure.

In this particular scenario, we erroneously return -EINVAL instead of the correct error code provided by the bdev_open_by_path() function, which is -EBUSY.

Fix this, by returning error code from the bdev_open_by_path() function. With this correction, the mount error message will align with that of ext4 and xfs.

---
 fs/btrfs/volumes.c | 17 ++++++++++++-----
 1 file changed, 12 insertions(+), 5 deletions(-)
fs/btrfs/volumes.c +12 -5
index 6fc2d99270c18..a9b44755fc6f0 100644
@@ -1233,25 +1233,32 @@ static int open_fs_devices(struct btrfs_fs_devices *fs_devices,
 	struct btrfs_device *device;
 	struct btrfs_device *latest_dev = NULL;
 	struct btrfs_device *tmp_device;
+	int ret = 0;
 
 	flags |= FMODE_EXCL;
 
 	list_for_each_entry_safe(device, tmp_device, &fs_devices->devices,
 				 dev_list) {
-		int ret;
+		int ret2;
 
-		ret = btrfs_open_one_device(fs_devices, device, flags, holder);
-		if (ret == 0 &&
+		ret2 = btrfs_open_one_device(fs_devices, device, flags, holder);
+		if (ret2 == 0 &&
 		    (!latest_dev || device->generation > latest_dev->generation)) {
 			latest_dev = device;
-		} else if (ret == -ENODATA) {
+		} else if (ret2 == -ENODATA) {
 			fs_devices->num_devices--;
 			list_del(&device->dev_list);
 			btrfs_free_device(device);
 		}
+		if (ret == 0 && ret2 != 0)
+			ret = ret2;
 	}
-	if (fs_devices->open_devices == 0)
+
+	if (fs_devices->open_devices == 0) {
+		if (ret)
+			return ret;
 		return -EINVAL;
+	}
 
 	fs_devices->opened = 1;
 	fs_devices->latest_dev = latest_dev;
Signature · 1 line
2.43.0
author maintainer
#11 lore raw

PATCH 12/13 bpf: Check bloom filter map value size

In reply to Sasha Levin

From: Andrei Matei <[email protected]>

[ Upstream commit a8d89feba7e54e691ca7c4efc2a6264fa83f3687 ]

This patch adds a missing check to bloom filter creating, rejecting values above KMALLOC_MAX_SIZE. This brings the bloom map in line with many other map types.

The lack of this protection can cause kernel crashes for value sizes that overflow int's. Such a crash was caught by syzkaller. The next patch adds more guard-rails at a lower level.

---
 kernel/bpf/bloom_filter.c                           | 13 +++++++++++++
 .../selftests/bpf/prog_tests/bloom_filter_map.c     |  6 ++++++
 2 files changed, 19 insertions(+)
kernel/bpf/bloom_filter.c +13 -0
index 48ee750849f25..78e810f49c445 100644
@@ -88,6 +88,18 @@ static int bloom_map_get_next_key(struct bpf_map *map, void *key, void *next_key
 	return -EOPNOTSUPP;
 }
 
+/* Called from syscall */
+static int bloom_map_alloc_check(union bpf_attr *attr)
+{
+	if (attr->value_size > KMALLOC_MAX_SIZE)
+		/* if value_size is bigger, the user space won't be able to
+		 * access the elements.
+		 */
+		return -E2BIG;
+
+	return 0;
+}
+
 static struct bpf_map *bloom_map_alloc(union bpf_attr *attr)
 {
 	u32 bitset_bytes, bitset_mask, nr_hash_funcs, nr_bits;
@@ -196,6 +208,7 @@ static int bloom_map_check_btf(const struct bpf_map *map,
 BTF_ID_LIST_SINGLE(bpf_bloom_map_btf_ids, struct, bpf_bloom_filter)
 const struct bpf_map_ops bloom_filter_map_ops = {
 	.map_meta_equal = bpf_map_meta_equal,
+	.map_alloc_check = bloom_map_alloc_check,
 	.map_alloc = bloom_map_alloc,
 	.map_free = bloom_map_free,
 	.map_get_next_key = bloom_map_get_next_key,
tools/testing/selftests/bpf/prog_tests/bloom_filter_map.c +6 -0
index d2d9e965eba59..f79815b7e951b 100644
@@ -2,6 +2,7 @@
 /* Copyright (c) 2021 Facebook */
 
 #include <sys/syscall.h>
+#include <limits.h>
 #include <test_progs.h>
 #include "bloom_filter_map.skel.h"
 
@@ -21,6 +22,11 @@ static void test_fail_cases(void)
 	if (!ASSERT_LT(fd, 0, "bpf_map_create bloom filter invalid value size 0"))
 		close(fd);
 
+	/* Invalid value size: too big */
+	fd = bpf_map_create(BPF_MAP_TYPE_BLOOM_FILTER, NULL, 0, INT32_MAX, 100, NULL);
+	if (!ASSERT_LT(fd, 0, "bpf_map_create bloom filter invalid value too large"))
+		close(fd);
+
 	/* Invalid max entries size */
 	fd = bpf_map_create(BPF_MAP_TYPE_BLOOM_FILTER, NULL, 0, sizeof(value), 0, NULL);
 	if (!ASSERT_LT(fd, 0, "bpf_map_create bloom filter invalid max entries size"))
Signature · 1 line
2.43.0
author maintainer
#12 lore raw

PATCH 13/13 kbuild: Disable KCSAN for autogenerated *.mod.c intermediaries

In reply to Sasha Levin

From: "Borislav Petkov (AMD)" <[email protected]>

[ Upstream commit 54babdc0343fff2f32dfaafaaa9e42c4db278204 ]

When KCSAN and CONSTRUCTORS are enabled, one can trigger the

"Unpatched return thunk in use. This should not happen!"

catch-all warning.

Usually, when objtool runs on the .o objects, it does generate a section .return_sites which contains all offsets in the objects to the return thunks of the functions present there. Those return thunks then get patched at runtime by the alternatives.

KCSAN and CONSTRUCTORS add this to the object file's .text.startup section:

-------------------
Disassembly of section .text.startup:

...

0000000000000010 <_sub_I_00099_0>:

    10:   f3 0f 1e fa             endbr64
    14:   e8 00 00 00 00          call   19 <_sub_I_00099_0+0x9>
                          15: R_X86_64_PLT32      __tsan_init-0x4
    19:   e9 00 00 00 00          jmp    1e <__UNIQUE_ID___addressable_cryptd_alloc_aead349+0x6>
                          1a: R_X86_64_PLT32      __x86_return_thunk-0x4

-------------------

which, if it is built as a module goes through the intermediary stage of creating a <module>.mod.c file which, when translated, receives a second constructor:

-------------------
Disassembly of section .text.startup:

0000000000000010 <_sub_I_00099_0>:

    10:   f3 0f 1e fa             endbr64
    14:   e8 00 00 00 00          call   19 <_sub_I_00099_0+0x9>
                          15: R_X86_64_PLT32      __tsan_init-0x4
    19:   e9 00 00 00 00          jmp    1e <_sub_I_00099_0+0xe>
                          1a: R_X86_64_PLT32      __x86_return_thunk-0x4

...

0000000000000030 <_sub_I_00099_0>:

    30:   f3 0f 1e fa             endbr64
    34:   e8 00 00 00 00          call   39 <_sub_I_00099_0+0x9>
                          35: R_X86_64_PLT32      __tsan_init-0x4
    39:   e9 00 00 00 00          jmp    3e <__ksymtab_cryptd_alloc_ahash+0x2>
                          3a: R_X86_64_PLT32      __x86_return_thunk-0x4

-------------------

in the .ko file.

Objtool has run already so that second constructor's return thunk cannot be added to the .return_sites section and thus the return thunk remains unpatched and the warning rightfully fires.

Drop KCSAN flags from the mod.c generation stage as those constructors do not contain data races one would be interested about.

Debugged together with David Kaplan <[email protected]> and Nikolay Borisov <[email protected]>.

---
 scripts/Makefile.modfinal | 2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)
scripts/Makefile.modfinal +1 -1
index 3af5e5807983a..650d59388336f 100644
@@ -23,7 +23,7 @@ modname = $(notdir $(@:.mod.o=))
 part-of-module = y
 
 quiet_cmd_cc_o_c = CC [M]  $@
-      cmd_cc_o_c = $(CC) $(filter-out $(CC_FLAGS_CFI) $(CFLAGS_GCOV), $(c_flags)) -c -o $@ $<
+      cmd_cc_o_c = $(CC) $(filter-out $(CC_FLAGS_CFI) $(CFLAGS_GCOV) $(CFLAGS_KCSAN), $(c_flags)) -c -o $@ $<
 
 %.mod.o: %.mod.c FORCE
 	$(call if_changed_dep,cc_o_c)
Signature · 1 line
2.43.0