Skip to contentSkip to conversation

dmaengine: ste_dma40: fix unneeded variable warning

PATCH In review Sasha Levin 2 participants 46 messages Aug 20, 2019

Subsystems (12)
lore
Share & tools
lore

Sasha Levin

author maintainer
#0 lore raw

PATCH 1/44 dmaengine: ste_dma40: fix unneeded variable warning

From: Arnd Bergmann <[email protected]>

[ Upstream commit 5d6fb560729a5d5554e23db8d00eb57cd0021083 ]

clang-9 points out that there are two variables that depending on the configuration may only be used in an ARRAY_SIZE() expression but not referenced:

drivers/dma/ste_dma40.c:145:12: error: variable 'd40_backup_regs' is not needed and will not be emitted [-Werror,-Wunneeded-internal-declaration]

static u32 d40_backup_regs[] = {
           ^

drivers/dma/ste_dma40.c:214:12: error: variable 'd40_backup_regs_chan' is not needed and will not be emitted [-Werror,-Wunneeded-internal-declaration] static u32 d40_backup_regs_chan[] = {

Mark these __maybe_unused to shut up the warning.

---
 drivers/dma/ste_dma40.c | 4 ++--
 1 file changed, 2 insertions(+), 2 deletions(-)
drivers/dma/ste_dma40.c +2 -2
index 89d710899010d..de8bfd9a76e9e 100644
@@ -142,7 +142,7 @@ enum d40_events {
  * when the DMA hw is powered off.
  * TODO: Add save/restore of D40_DREG_GCC on dma40 v3 or later, if that works.
  */
-static u32 d40_backup_regs[] = {
+static __maybe_unused u32 d40_backup_regs[] = {
 	D40_DREG_LCPA,
 	D40_DREG_LCLA,
 	D40_DREG_PRMSE,
@@ -211,7 +211,7 @@ static u32 d40_backup_regs_v4b[] = {
 
 #define BACKUP_REGS_SZ_V4B ARRAY_SIZE(d40_backup_regs_v4b)
 
-static u32 d40_backup_regs_chan[] = {
+static __maybe_unused u32 d40_backup_regs_chan[] = {
 	D40_CHAN_REG_SSCFG,
 	D40_CHAN_REG_SSELT,
 	D40_CHAN_REG_SSPTR,
Signature · 1 line
2.20.1
author maintainer
#1 lore raw

PATCH 2/44 nvme-multipath: revalidate nvme_ns_head gendisk in nvme_validate_ns

From: Anthony Iliopoulos <[email protected]>

[ Upstream commit fab7772bfbcfe8fb8e3e352a6a8fcaf044cded17 ]

When CONFIG_NVME_MULTIPATH is set, only the hidden gendisk associated with the per-controller ns is run through revalidate_disk when a rescan is triggered, while the visible blockdev never gets its size (bdev->bd_inode->i_size) updated to reflect any capacity changes that may have occurred.

This prevents online resizing of nvme block devices and in extension of any filesystems atop that will are unable to expand while mounted, as userspace relies on the blockdev size for obtaining the disk capacity (via BLKGETSIZE/64 ioctls).

Fix this by explicitly revalidating the actual namespace gendisk in addition to the per-controller gendisk, when multipath is enabled.

---
 drivers/nvme/host/core.c | 1 +
 1 file changed, 1 insertion(+)
drivers/nvme/host/core.c +1 -0
index 5deb4deb38209..ab1a2b1ec3637 100644
@@ -1668,6 +1668,7 @@ static void __nvme_revalidate_disk(struct gendisk *disk, struct nvme_id_ns *id)
 	if (ns->head->disk) {
 		nvme_update_disk_info(ns->head->disk, ns, id);
 		blk_queue_stack_limits(ns->head->disk->queue, ns->queue);
+		revalidate_disk(ns->head->disk);
 	}
 #endif
 }
Signature · 1 line
2.20.1
author maintainer
#2 lore raw

PATCH 3/44 afs: Fix the CB.ProbeUuid service handler to reply correctly

In reply to Sasha Levin

From: David Howells <[email protected]>

[ Upstream commit 2067b2b3f4846402a040286135f98f46f8919939 ]

Fix the service handler function for the CB.ProbeUuid RPC call so that it replies in the correct manner - that is an empty reply for success and an abort of 1 for failure.

Putting 0 or 1 in an integer in the body of the reply should result in the fileserver throwing an RX_PROTOCOL_ERROR abort and discarding its record of the client; older servers, however, don't necessarily check that all the data got consumed, and so might incorrectly think that they got a positive response and associate the client with the wrong host record.

If the client is incorrectly associated, this will result in callbacks intended for a different client being delivered to this one and then, when the other client connects and responds positively, all of the callback promises meant for the client that issued the improper response will be lost and it won't receive any further change notifications.

Fixes 9396d496d745 ("afs: support the CB.ProbeUuid RPC op")
---
 fs/afs/cmservice.c | 10 +++-------
 1 file changed, 3 insertions(+), 7 deletions(-)
fs/afs/cmservice.c +3 -7
index 3451be03667f0..00033a481ba05 100644
@@ -502,18 +502,14 @@ static void SRXAFSCB_ProbeUuid(struct work_struct *work)
 	struct afs_call *call = container_of(work, struct afs_call, work);
 	struct afs_uuid *r = call->request;
 
-	struct {
-		__be32	match;
-	} reply;
-
 	_enter("");
 
 	if (memcmp(r, &call->net->uuid, sizeof(call->net->uuid)) == 0)
-		reply.match = htonl(0);
+		afs_send_empty_reply(call);
 	else
-		reply.match = htonl(1);
+		rxrpc_kernel_abort_call(call->net->socket, call->rxcall,
+					1, 1, "K-1");
 
-	afs_send_simple_reply(call, &reply, sizeof(reply));
 	afs_put_call(call);
 	_leave("");
 }
Signature · 1 line
2.20.1
author maintainer
#3 lore raw

PATCH 4/44 afs: Fix loop index mixup in afs_deliver_vl_get_entry_by_name_u()

In reply to Sasha Levin

From: Marc Dionne <[email protected]>

[ Upstream commit 4a46fdba449a5cd890271df5a9e23927d519ed00 ]

afs_deliver_vl_get_entry_by_name_u() scans through the vl entry received from the volume location server and builds a return list containing the sites that are currently valid. When assigning values for the return list, the index into the vl entry (i) is used rather than the one for the new list (entry->nr_server). If all sites are usable, this works out fine as the indices will match. If some sites are not valid, for example if AFS_VLSF_DONTUSE is set, fs_mask and the uuid will be set for the wrong return site.

Fix this by using entry->nr_server as the index into the arrays being filled in rather than i.

This can lead to EDESTADDRREQ errors if none of the returned sites have a valid fs_mask.

Fixes d2ddc776a458 ("afs: Overhaul volume and server record caching and fileserver rotation")
---
 fs/afs/vlclient.c | 11 ++++++-----
 1 file changed, 6 insertions(+), 5 deletions(-)
fs/afs/vlclient.c +6 -5
index d7e0fd3c00df9..cfb0ac4bd039e 100644
@@ -56,23 +56,24 @@ static int afs_deliver_vl_get_entry_by_name_u(struct afs_call *call)
 		struct afs_uuid__xdr *xdr;
 		struct afs_uuid *uuid;
 		int j;
+		int n = entry->nr_servers;
 
 		tmp = ntohl(uvldb->serverFlags[i]);
 		if (tmp & AFS_VLSF_DONTUSE ||
 		    (new_only && !(tmp & AFS_VLSF_NEWREPSITE)))
 			continue;
 		if (tmp & AFS_VLSF_RWVOL) {
-			entry->fs_mask[i] |= AFS_VOL_VTM_RW;
+			entry->fs_mask[n] |= AFS_VOL_VTM_RW;
 			if (vlflags & AFS_VLF_BACKEXISTS)
-				entry->fs_mask[i] |= AFS_VOL_VTM_BAK;
+				entry->fs_mask[n] |= AFS_VOL_VTM_BAK;
 		}
 		if (tmp & AFS_VLSF_ROVOL)
-			entry->fs_mask[i] |= AFS_VOL_VTM_RO;
-		if (!entry->fs_mask[i])
+			entry->fs_mask[n] |= AFS_VOL_VTM_RO;
+		if (!entry->fs_mask[n])
 			continue;
 
 		xdr = &uvldb->serverNumber[i];
-		uuid = (struct afs_uuid *)&entry->fs_server[i];
+		uuid = (struct afs_uuid *)&entry->fs_server[n];
 		uuid->time_low			= xdr->time_low;
 		uuid->time_mid			= htons(ntohl(xdr->time_mid));
 		uuid->time_hi_and_version	= htons(ntohl(xdr->time_hi_and_version));
Signature · 1 line
2.20.1
author maintainer
#4 lore raw

PATCH 5/44 fs: afs: Fix a possible null-pointer dereference in afs_put_read()

In reply to Sasha Levin

From: Jia-Ju Bai <[email protected]>

[ Upstream commit a6eed4ab5dd4bfb696c1a3f49742b8d1846a66a0 ]

In afs_read_dir(), there is an if statement on line 255 to check whether req->pages is NULL:

	if (!req->pages)
		goto error;

If req->pages is NULL, afs_put_read() on line 337 is executed. In afs_put_read(), req->pages[i] is used on line 195. Thus, a possible null-pointer dereference may occur in this case.

To fix this possible bug, an if statement is added in afs_put_read() to check req->pages.

This bug is found by a static analysis tool STCheck written by us.

Fixes f3ddee8dc4e2 ("afs: Fix directory handling")
---
 fs/afs/file.c | 12 +++++++-----
 1 file changed, 7 insertions(+), 5 deletions(-)
fs/afs/file.c +7 -5
index 8fd7d3b9a1b1f..87beabc7114ee 100644
@@ -191,11 +191,13 @@ void afs_put_read(struct afs_read *req)
 	int i;
 
 	if (refcount_dec_and_test(&req->usage)) {
-		for (i = 0; i < req->nr_pages; i++)
-			if (req->pages[i])
-				put_page(req->pages[i]);
-		if (req->pages != req->array)
-			kfree(req->pages);
+		if (req->pages) {
+			for (i = 0; i < req->nr_pages; i++)
+				if (req->pages[i])
+					put_page(req->pages[i]);
+			if (req->pages != req->array)
+				kfree(req->pages);
+		}
 		kfree(req);
 	}
 }
Signature · 1 line
2.20.1
author maintainer
#5 lore raw

PATCH 6/44 afs: Fix off-by-one in afs_rename() expected data version calculation

In reply to Sasha Levin

From: David Howells <[email protected]>

[ Upstream commit 37c0bbb3326674940e657118306ac52364314523 ]

When afs_rename() calculates the expected data version of the target directory in a cross-directory rename, it doesn't increment it as it should, so it always thinks that the target inode is unexpectedly modified on the server.

Fixes a58823ac4589 ("afs: Fix application of status and callback to be under same lock")
---
 fs/afs/dir.c | 2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)
fs/afs/dir.c +1 -1
index da9563d62b327..9750ac70f8ffb 100644
@@ -1807,7 +1807,7 @@ static int afs_rename(struct inode *old_dir, struct dentry *old_dentry,
 				afs_end_vnode_operation(&fc);
 				goto error_rehash;
 			}
-			new_data_version = new_dvnode->status.data_version;
+			new_data_version = new_dvnode->status.data_version + 1;
 		} else {
 			new_data_version = orig_data_version;
 			new_scb = &scb[0];
Signature · 1 line
2.20.1
author maintainer
#6 lore raw

PATCH 7/44 afs: Only update d_fsdata if different in afs_d_revalidate()

In reply to Sasha Levin

From: David Howells <[email protected]>

[ Upstream commit 5dc84855b0fc7e1db182b55c5564fd539d6eff92 ]

In the in-kernel afs filesystem, d_fsdata is set with the data version of the parent directory. afs_d_revalidate() will update this to the current directory version, but it shouldn't do this if it the value it read from d_fsdata is the same as no lock is held and cmpxchg() is not used.

Fix the code to only change the value if it is different from the current directory version.

Fixes 260a980317da ("[AFS]: Add "directory write" support.")
---
 fs/afs/dir.c | 3 ++-
 1 file changed, 2 insertions(+), 1 deletion(-)
fs/afs/dir.c +2 -1
index 9750ac70f8ffb..b87b41721eaa8 100644
@@ -1018,7 +1018,7 @@ static int afs_d_revalidate(struct dentry *dentry, unsigned int flags)
 	dir_version = (long)dir->status.data_version;
 	de_version = (long)dentry->d_fsdata;
 	if (de_version == dir_version)
-		goto out_valid;
+		goto out_valid_noupdate;
 
 	dir_version = (long)dir->invalid_before;
 	if (de_version - dir_version >= 0)
@@ -1082,6 +1082,7 @@ static int afs_d_revalidate(struct dentry *dentry, unsigned int flags)
 
 out_valid:
 	dentry->d_fsdata = (void *)dir_version;
+out_valid_noupdate:
 	dput(parent);
 	key_put(key);
 	_leave(" = 1 [valid]");
Signature · 1 line
2.20.1
author maintainer
#7 lore raw

PATCH 8/44 afs: Fix missing dentry data version updating

In reply to Sasha Levin

From: David Howells <[email protected]>

[ Upstream commit 9dd0b82ef530cdfe805c9f7079c99e104be59a14 ]

In the in-kernel afs filesystem, the d_fsdata dentry field is used to hold the data version of the parent directory when it was created or when d_revalidate() last caused it to be updated. This is compared to the ->invalid_before field in the directory inode, rather than the actual data version number, thereby allowing changes due to local edits to be ignored. Only if the server data version gets bumped unexpectedly (eg. by a competing client), do we need to revalidate stuff.

However, the d_fsdata field should also be updated if an rpc op is performed that modifies that particular dentry. Such ops return the revised data version of the directory(ies) involved, so we should use that.

This is particularly problematic for rename, since a dentry from one directory may be moved directly into another directory (ie. mv a/x b/x). It would then be sporting the wrong data version - and if this is in the future, for the destination directory, revalidations would be missed, leading to foreign renames and hard-link deletion being missed.

Fix this by the following means:

(1) Return the data version number from operations that read the directory contents - if they issue the read. This starts in afs_dir_iterate() and is used, ignored or passed back by its callers.

(2) In afs_lookup*(), set the dentry version to the version returned by
(1) before d_splice_alias() is called and the dentry published.

(3) In afs_d_revalidate(), set the dentry version to that returned from
(1) if an rpc call was issued. This means that if a parallel procedure, such as mkdir(), modifies the directory, we won't accidentally use the data version from that.

(4) In afs_{mkdir,create,link,symlink}(), set the new dentry's version to the directory data version before d_instantiate() is called.

(5) In afs_{rmdir,unlink}, update the target dentry's version to the directory data version as soon as we've updated the directory inode.

(6) In afs_rename(), we need to unhash the old dentry before we start so that we don't get afs_d_revalidate() reverting the version change in cross-directory renames.

We then need to set both the old and the new dentry versions the data version of the new directory before we call d_move() as d_move() will rehash them.

Fixes 1da177e4c3f4 ("Linux-2.6.12-rc2")
---
 fs/afs/dir.c | 84 +++++++++++++++++++++++++++++++++++++++++++---------
 1 file changed, 70 insertions(+), 14 deletions(-)
fs/afs/dir.c +70 -14
index b87b41721eaa8..9620f19308f58 100644
@@ -441,7 +441,7 @@ static int afs_dir_iterate_block(struct afs_vnode *dvnode,
  * iterate through the data blob that lists the contents of an AFS directory
  */
 static int afs_dir_iterate(struct inode *dir, struct dir_context *ctx,
-			   struct key *key)
+			   struct key *key, afs_dataversion_t *_dir_version)
 {
 	struct afs_vnode *dvnode = AFS_FS_I(dir);
 	struct afs_xdr_dir_page *dbuf;
@@ -461,6 +461,7 @@ static int afs_dir_iterate(struct inode *dir, struct dir_context *ctx,
 	req = afs_read_dir(dvnode, key);
 	if (IS_ERR(req))
 		return PTR_ERR(req);
+	*_dir_version = req->data_version;
 
 	/* round the file position up to the next entry boundary */
 	ctx->pos += sizeof(union afs_xdr_dirent) - 1;
@@ -515,7 +516,10 @@ static int afs_dir_iterate(struct inode *dir, struct dir_context *ctx,
  */
 static int afs_readdir(struct file *file, struct dir_context *ctx)
 {
-	return afs_dir_iterate(file_inode(file), ctx, afs_file_key(file));
+	afs_dataversion_t dir_version;
+
+	return afs_dir_iterate(file_inode(file), ctx, afs_file_key(file),
+			       &dir_version);
 }
 
 /*
@@ -556,7 +560,8 @@ static int afs_lookup_one_filldir(struct dir_context *ctx, const char *name,
  * - just returns the FID the dentry name maps to if found
  */
 static int afs_do_lookup_one(struct inode *dir, struct dentry *dentry,
-			     struct afs_fid *fid, struct key *key)
+			     struct afs_fid *fid, struct key *key,
+			     afs_dataversion_t *_dir_version)
 {
 	struct afs_super_info *as = dir->i_sb->s_fs_info;
 	struct afs_lookup_one_cookie cookie = {
@@ -569,7 +574,7 @@ static int afs_do_lookup_one(struct inode *dir, struct dentry *dentry,
 	_enter("{%lu},%p{%pd},", dir->i_ino, dentry, dentry);
 
 	/* search the directory */
-	ret = afs_dir_iterate(dir, &cookie.ctx, key);
+	ret = afs_dir_iterate(dir, &cookie.ctx, key, _dir_version);
 	if (ret < 0) {
 		_leave(" = %d [iter]", ret);
 		return ret;
@@ -643,6 +648,7 @@ static struct inode *afs_do_lookup(struct inode *dir, struct dentry *dentry,
 	struct afs_server *server;
 	struct afs_vnode *dvnode = AFS_FS_I(dir), *vnode;
 	struct inode *inode = NULL, *ti;
+	afs_dataversion_t data_version = READ_ONCE(dvnode->status.data_version);
 	int ret, i;
 
 	_enter("{%lu},%p{%pd},", dir->i_ino, dentry, dentry);
@@ -670,12 +676,14 @@ static struct inode *afs_do_lookup(struct inode *dir, struct dentry *dentry,
 		cookie->fids[i].vid = as->volume->vid;
 
 	/* search the directory */
-	ret = afs_dir_iterate(dir, &cookie->ctx, key);
+	ret = afs_dir_iterate(dir, &cookie->ctx, key, &data_version);
 	if (ret < 0) {
 		inode = ERR_PTR(ret);
 		goto out;
 	}
 
+	dentry->d_fsdata = (void *)(unsigned long)data_version;
+
 	inode = ERR_PTR(-ENOENT);
 	if (!cookie->found)
 		goto out;
@@ -969,7 +977,8 @@ static int afs_d_revalidate(struct dentry *dentry, unsigned int flags)
 	struct dentry *parent;
 	struct inode *inode;
 	struct key *key;
-	long dir_version, de_version;
+	afs_dataversion_t dir_version;
+	long de_version;
 	int ret;
 
 	if (flags & LOOKUP_RCU)
@@ -1015,20 +1024,20 @@ static int afs_d_revalidate(struct dentry *dentry, unsigned int flags)
 	 * on a 32-bit system, we only have 32 bits in the dentry to store the
 	 * version.
 	 */
-	dir_version = (long)dir->status.data_version;
+	dir_version = dir->status.data_version;
 	de_version = (long)dentry->d_fsdata;
-	if (de_version == dir_version)
+	if (de_version == (long)dir_version)
 		goto out_valid_noupdate;
 
-	dir_version = (long)dir->invalid_before;
-	if (de_version - dir_version >= 0)
+	dir_version = dir->invalid_before;
+	if (de_version - (long)dir_version >= 0)
 		goto out_valid;
 
 	_debug("dir modified");
 	afs_stat_v(dir, n_reval);
 
 	/* search the directory for this vnode */
-	ret = afs_do_lookup_one(&dir->vfs_inode, dentry, &fid, key);
+	ret = afs_do_lookup_one(&dir->vfs_inode, dentry, &fid, key, &dir_version);
 	switch (ret) {
 	case 0:
 		/* the filename maps to something */
@@ -1081,7 +1090,7 @@ static int afs_d_revalidate(struct dentry *dentry, unsigned int flags)
 	}
 
 out_valid:
-	dentry->d_fsdata = (void *)dir_version;
+	dentry->d_fsdata = (void *)(unsigned long)dir_version;
 out_valid_noupdate:
 	dput(parent);
 	key_put(key);
@@ -1187,6 +1196,20 @@ static void afs_prep_for_new_inode(struct afs_fs_cursor *fc,
 	iget_data->cb_s_break = fc->cbi->server->cb_s_break;
 }
 
+/*
+ * Note that a dentry got changed.  We need to set d_fsdata to the data version
+ * number derived from the result of the operation.  It doesn't matter if
+ * d_fsdata goes backwards as we'll just revalidate.
+ */
+static void afs_update_dentry_version(struct afs_fs_cursor *fc,
+				      struct dentry *dentry,
+				      struct afs_status_cb *scb)
+{
+	if (fc->ac.error == 0)
+		dentry->d_fsdata =
+			(void *)(unsigned long)scb->status.data_version;
+}
+
 /*
  * create a directory on an AFS filesystem
  */
@@ -1229,6 +1252,7 @@ static int afs_mkdir(struct inode *dir, struct dentry *dentry, umode_t mode)
 		afs_check_for_remote_deletion(&fc, dvnode);
 		afs_vnode_commit_status(&fc, dvnode, fc.cb_break,
 					&data_version, &scb[0]);
+		afs_update_dentry_version(&fc, dentry, &scb[0]);
 		afs_vnode_new_inode(&fc, dentry, &iget_data, &scb[1]);
 		ret = afs_end_vnode_operation(&fc);
 		if (ret < 0)
@@ -1321,6 +1345,7 @@ static int afs_rmdir(struct inode *dir, struct dentry *dentry)
 
 		afs_vnode_commit_status(&fc, dvnode, fc.cb_break,
 					&data_version, scb);
+		afs_update_dentry_version(&fc, dentry, scb);
 		ret = afs_end_vnode_operation(&fc);
 		if (ret == 0) {
 			afs_dir_remove_subdir(dentry);
@@ -1462,6 +1487,7 @@ static int afs_unlink(struct inode *dir, struct dentry *dentry)
 					&data_version, &scb[0]);
 		afs_vnode_commit_status(&fc, vnode, fc.cb_break_2,
 					&data_version_2, &scb[1]);
+		afs_update_dentry_version(&fc, dentry, &scb[0]);
 		ret = afs_end_vnode_operation(&fc);
 		if (ret == 0 && !(scb[1].have_status || scb[1].have_error))
 			ret = afs_dir_remove_link(dvnode, dentry, key);
@@ -1530,6 +1556,7 @@ static int afs_create(struct inode *dir, struct dentry *dentry, umode_t mode,
 		afs_check_for_remote_deletion(&fc, dvnode);
 		afs_vnode_commit_status(&fc, dvnode, fc.cb_break,
 					&data_version, &scb[0]);
+		afs_update_dentry_version(&fc, dentry, &scb[0]);
 		afs_vnode_new_inode(&fc, dentry, &iget_data, &scb[1]);
 		ret = afs_end_vnode_operation(&fc);
 		if (ret < 0)
@@ -1611,6 +1638,7 @@ static int afs_link(struct dentry *from, struct inode *dir,
 		afs_vnode_commit_status(&fc, vnode, fc.cb_break_2,
 					NULL, &scb[1]);
 		ihold(&vnode->vfs_inode);
+		afs_update_dentry_version(&fc, dentry, &scb[0]);
 		d_instantiate(dentry, &vnode->vfs_inode);
 
 		mutex_unlock(&vnode->io_lock);
@@ -1690,6 +1718,7 @@ static int afs_symlink(struct inode *dir, struct dentry *dentry,
 		afs_check_for_remote_deletion(&fc, dvnode);
 		afs_vnode_commit_status(&fc, dvnode, fc.cb_break,
 					&data_version, &scb[0]);
+		afs_update_dentry_version(&fc, dentry, &scb[0]);
 		afs_vnode_new_inode(&fc, dentry, &iget_data, &scb[1]);
 		ret = afs_end_vnode_operation(&fc);
 		if (ret < 0)
@@ -1795,6 +1824,17 @@ static int afs_rename(struct inode *old_dir, struct dentry *old_dentry,
 		}
 	}
 
+	/* This bit is potentially nasty as there's a potential race with
+	 * afs_d_revalidate{,_rcu}().  We have to change d_fsdata on the dentry
+	 * to reflect it's new parent's new data_version after the op, but
+	 * d_revalidate may see old_dentry between the op having taken place
+	 * and the version being updated.
+	 *
+	 * So drop the old_dentry for now to make other threads go through
+	 * lookup instead - which we hold a lock against.
+	 */
+	d_drop(old_dentry);
+
 	ret = -ERESTARTSYS;
 	if (afs_begin_vnode_operation(&fc, orig_dvnode, key, true)) {
 		afs_dataversion_t orig_data_version;
@@ -1806,7 +1846,7 @@ static int afs_rename(struct inode *old_dir, struct dentry *old_dentry,
 		if (orig_dvnode != new_dvnode) {
 			if (mutex_lock_interruptible_nested(&new_dvnode->io_lock, 1) < 0) {
 				afs_end_vnode_operation(&fc);
-				goto error_rehash;
+				goto error_rehash_old;
 			}
 			new_data_version = new_dvnode->status.data_version + 1;
 		} else {
@@ -1831,7 +1871,7 @@ static int afs_rename(struct inode *old_dir, struct dentry *old_dentry,
 		}
 		ret = afs_end_vnode_operation(&fc);
 		if (ret < 0)
-			goto error_rehash;
+			goto error_rehash_old;
 	}
 
 	if (ret == 0) {
@@ -1857,10 +1897,26 @@ static int afs_rename(struct inode *old_dir, struct dentry *old_dentry,
 				drop_nlink(new_inode);
 			spin_unlock(&new_inode->i_lock);
 		}
+
+		/* Now we can update d_fsdata on the dentries to reflect their
+		 * new parent's data_version.
+		 *
+		 * Note that if we ever implement RENAME_EXCHANGE, we'll have
+		 * to update both dentries with opposing dir versions.
+		 */
+		if (new_dvnode != orig_dvnode) {
+			afs_update_dentry_version(&fc, old_dentry, &scb[1]);
+			afs_update_dentry_version(&fc, new_dentry, &scb[1]);
+		} else {
+			afs_update_dentry_version(&fc, old_dentry, &scb[0]);
+			afs_update_dentry_version(&fc, new_dentry, &scb[0]);
+		}
 		d_move(old_dentry, new_dentry);
 		goto error_tmp;
 	}
 
+error_rehash_old:
+	d_rehash(new_dentry);
 error_rehash:
 	if (rehash)
 		d_rehash(rehash);
Signature · 1 line
2.20.1
author maintainer
#8 lore raw

PATCH 9/44 intel_th: Use the correct style for SPDX License Identifier

In reply to Sasha Levin

From: Nishad Kamdar <[email protected]>

[ Upstream commit fac7b714c514fcc555541e1d6450c694b0a5f8d3 ]

This patch corrects the SPDX License Identifier style in header files related to Drivers for Intel(R) Trace Hub controller.
For C header files Documentation/process/license-rules.rst mandates C-like comments (opposed to C source files where C++ style should be used)

Changes made by using a script provided by Joe Perches here:
https://lkml.org/lkml/2019/2/7/46

---
 drivers/hwtracing/intel_th/msu.h | 2 +-
 drivers/hwtracing/intel_th/pti.h | 2 +-
 2 files changed, 2 insertions(+), 2 deletions(-)
drivers/hwtracing/intel_th/msu.h +1 -1
index 574c16004cb2e..13d9b141daaaa 100644
@@ -1,4 +1,4 @@
-// SPDX-License-Identifier: GPL-2.0
+/* SPDX-License-Identifier: GPL-2.0 */
 /*
  * Intel(R) Trace Hub Memory Storage Unit (MSU) data structures
  *
drivers/hwtracing/intel_th/pti.h +1 -1
index e9381babc84c3..7dfc0431333b7 100644
@@ -1,4 +1,4 @@
-// SPDX-License-Identifier: GPL-2.0
+/* SPDX-License-Identifier: GPL-2.0 */
 /*
  * Intel(R) Trace Hub PTI output data structures
  *
Signature · 1 line
2.20.1
#9 lore raw

Re: [PATCH AUTOSEL 5.2 09/44] intel_th: Use the correct style for SPDX License Identifier

Quoted 14 lines · Sasha Levin

From: Nishad Kamdar <[email protected]>

[ Upstream commit fac7b714c514fcc555541e1d6450c694b0a5f8d3 ]

This patch corrects the SPDX License Identifier style in header files related to Drivers for Intel(R) Trace Hub controller.
For C header files Documentation/process/license-rules.rst mandates C-like comments (opposed to C source files where C++ style should be used)

Changes made by using a script provided by Joe Perches here:
https://lkml.org/lkml/2019/2/7/46

Quoted 8 lines
---
 drivers/hwtracing/intel_th/msu.h | 2 +-
 drivers/hwtracing/intel_th/pti.h | 2 +-
 2 files changed, 2 insertions(+), 2 deletions(-)

Not really a stable patch at all, unless you want to start backporting all SPDX changes (hint, NO we do not!) :)

please drop this from everywhere.

And what triggered this? It's just comment changes, shouldn't the autobot know to ignore those?

thanks,

greg k-h

author maintainer
#10 lore raw

Re: [PATCH AUTOSEL 5.2 09/44] intel_th: Use the correct style for SPDX License Identifier

Quoted 24 lines · Greg Kroah-Hartman
Quoted 14 lines · Sasha Levin

From: Nishad Kamdar <[email protected]>

[ Upstream commit fac7b714c514fcc555541e1d6450c694b0a5f8d3 ]

This patch corrects the SPDX License Identifier style in header files related to Drivers for Intel(R) Trace Hub controller.
For C header files Documentation/process/license-rules.rst mandates C-like comments (opposed to C source files where C++ style should be used)

Changes made by using a script provided by Joe Perches here:
https://lkml.org/lkml/2019/2/7/46

Quoted 8 lines
---
 drivers/hwtracing/intel_th/msu.h | 2 +-
 drivers/hwtracing/intel_th/pti.h | 2 +-
 2 files changed, 2 insertions(+), 2 deletions(-)
Quoted 7 lines

Not really a stable patch at all, unless you want to start backporting all SPDX changes (hint, NO we do not!) :)

please drop this from everywhere.

And what triggered this? It's just comment changes, shouldn't the autobot know to ignore those?

It got a score just above my cutoff, and I missed it during review :(

FWIW, there's no explicit rule to ignore documentation, AUTOSEL sort of learned to do it on it's own, but here it seems that the content of the commit message outweighed the code change metrics.

--
Thanks,
Sasha

author maintainer
#11 lore raw

PATCH 10/44 nvmet: Fix use-after-free bug when a port is removed

In reply to Sasha Levin

From: Logan Gunthorpe <[email protected]>

[ Upstream commit 3aed86731ee2b23e4dc4d2c6d943d33992cd551b ]

When a port is removed through configfs, any connected controllers are still active and can still send commands. This causes a use-after-free bug which is detected by KASAN for any admin command that dereferences req->port (like in nvmet_execute_identify_ctrl).

To fix this, disconnect all active controllers when a subsystem is removed from a port. This ensures there are no active controllers when the port is eventually removed.

---
 drivers/nvme/target/configfs.c |  1 +
 drivers/nvme/target/core.c     | 12 ++++++++++++
 drivers/nvme/target/nvmet.h    |  3 +++
 3 files changed, 16 insertions(+)
drivers/nvme/target/configfs.c +1 -0
index 08dd5af357f7c..3854363118ccf 100644
@@ -673,6 +673,7 @@ static void nvmet_port_subsys_drop_link(struct config_item *parent,
 
 found:
 	list_del(&p->entry);
+	nvmet_port_del_ctrls(port, subsys);
 	nvmet_port_disc_changed(port, subsys);
 
 	if (list_empty(&port->subsystems))
drivers/nvme/target/core.c +12 -0
index 7734a6acff851..e4db9a4411681 100644
@@ -277,6 +277,18 @@ void nvmet_unregister_transport(const struct nvmet_fabrics_ops *ops)
 }
 EXPORT_SYMBOL_GPL(nvmet_unregister_transport);
 
+void nvmet_port_del_ctrls(struct nvmet_port *port, struct nvmet_subsys *subsys)
+{
+	struct nvmet_ctrl *ctrl;
+
+	mutex_lock(&subsys->lock);
+	list_for_each_entry(ctrl, &subsys->ctrls, subsys_entry) {
+		if (ctrl->port == port)
+			ctrl->ops->delete_ctrl(ctrl);
+	}
+	mutex_unlock(&subsys->lock);
+}
+
 int nvmet_enable_port(struct nvmet_port *port)
 {
 	const struct nvmet_fabrics_ops *ops;
drivers/nvme/target/nvmet.h +3 -0
index c25d88fc9dec8..b6b0d483e0c50 100644
@@ -415,6 +415,9 @@ void nvmet_port_send_ana_event(struct nvmet_port *port);
 int nvmet_register_transport(const struct nvmet_fabrics_ops *ops);
 void nvmet_unregister_transport(const struct nvmet_fabrics_ops *ops);
 
+void nvmet_port_del_ctrls(struct nvmet_port *port,
+			  struct nvmet_subsys *subsys);
+
 int nvmet_enable_port(struct nvmet_port *port);
 void nvmet_disable_port(struct nvmet_port *port);
 
Signature · 1 line
2.20.1
author maintainer
#12 lore raw

PATCH 11/44 nvmet-loop: Flush nvme_delete_wq when removing the port

In reply to Sasha Levin

From: Logan Gunthorpe <[email protected]>

[ Upstream commit 86b9a63e595ff03f9d0a7b92b6acc231fecefc29 ]

After calling nvme_loop_delete_ctrl(), the controllers will not yet be deleted because nvme_delete_ctrl() only schedules work to do the delete.

This means a race can occur if a port is removed but there are still active controllers trying to access that memory.

To fix this, flush the nvme_delete_wq before returning from nvme_loop_remove_port() so that any controllers that might be in the process of being deleted won't access a freed port.

---
 drivers/nvme/target/loop.c | 8 ++++++++
 1 file changed, 8 insertions(+)
drivers/nvme/target/loop.c +8 -0
index 9e211ad6bdd3d..da9cd07461fbb 100644
@@ -654,6 +654,14 @@ static void nvme_loop_remove_port(struct nvmet_port *port)
 	mutex_lock(&nvme_loop_ports_mutex);
 	list_del_init(&port->entry);
 	mutex_unlock(&nvme_loop_ports_mutex);
+
+	/*
+	 * Ensure any ctrls that are in the process of being
+	 * deleted are in fact deleted before we return
+	 * and free the port. This is to prevent active
+	 * ctrls from using a port after it's freed.
+	 */
+	flush_workqueue(nvme_delete_wq);
 }
 
 static const struct nvmet_fabrics_ops nvme_loop_ops = {
Signature · 1 line
2.20.1
author maintainer
#13 lore raw

PATCH 12/44 nvmet-file: fix nvmet_file_flush() always returning an error

In reply to Sasha Levin

From: Logan Gunthorpe <[email protected]>

[ Upstream commit cfc1a1af56200362d1508b82b9a3cc3acb2eae0c ]

Presently, nvmet_file_flush() always returns a call to errno_to_nvme_status() but that helper doesn't take into account the case when errno=0. So nvmet_file_flush() always returns an error code.

All other callers of errno_to_nvme_status() check for success before calling it.

To fix this, ensure errno_to_nvme_status() returns success if the errno is zero. This should prevent future mistakes like this from happening.

Fixes c6aa3542e010 ("nvmet: add error log support for file backend")
---
 drivers/nvme/target/core.c | 3 +++
 1 file changed, 3 insertions(+)
drivers/nvme/target/core.c +3 -0
index e4db9a4411681..396cbc7ea3532 100644
@@ -43,6 +43,9 @@ inline u16 errno_to_nvme_status(struct nvmet_req *req, int errno)
 	u16 status;
 
 	switch (errno) {
+	case 0:
+		status = NVME_SC_SUCCESS;
+		break;
 	case -ENOSPC:
 		req->error_loc = offsetof(struct nvme_rw_command, length);
 		status = NVME_SC_CAP_EXCEEDED | NVME_SC_DNR;
Signature · 1 line
2.20.1
author maintainer
#14 lore raw

PATCH 13/44 nvme-core: Fix extra device_put() call on error path

In reply to Sasha Levin

From: Logan Gunthorpe <[email protected]>

[ Upstream commit 8c36e66fb407ce076535a7db98ab9f6d720b866a ]

In the error path for nvme_init_subsystem(), nvme_put_subsystem() will call device_put(), but it will get called again after the mutex_unlock().

The device_put() only needs to be called if device_add() fails.

This bug caused a KASAN use-after-free error when adding and removing subsytems in a loop:

  BUG: KASAN: use-after-free in device_del+0x8d9/0x9a0
  Read of size 8 at addr ffff8883cdaf7120 by task multipathd/329

  CPU: 0 PID: 329 Comm: multipathd Not tainted 5.2.0-rc6-vmlocalyes-00019-g70a2b39005fd-dirty #314
  Hardware name: QEMU Standard PC (i440FX + PIIX, 1996), BIOS 1.10.2-1 04/01/2014
  Call Trace:
   dump_stack+0x7b/0xb5
   print_address_description+0x6f/0x280
   ? device_del+0x8d9/0x9a0
   __kasan_report+0x148/0x199
   ? device_del+0x8d9/0x9a0
   ? class_release+0x100/0x130
   ? device_del+0x8d9/0x9a0
   kasan_report+0x12/0x20
   __asan_report_load8_noabort+0x14/0x20
   device_del+0x8d9/0x9a0
   ? device_platform_notify+0x70/0x70
   nvme_destroy_subsystem+0xf9/0x150
   nvme_free_ctrl+0x280/0x3a0
   device_release+0x72/0x1d0
   kobject_put+0x144/0x410
   put_device+0x13/0x20
   nvme_free_ns+0xc4/0x100
   nvme_release+0xb3/0xe0
   __blkdev_put+0x549/0x6e0
   ? kasan_check_write+0x14/0x20
   ? bd_set_size+0xb0/0xb0
   ? kasan_check_write+0x14/0x20
   ? mutex_lock+0x8f/0xe0
   ? __mutex_lock_slowpath+0x20/0x20
   ? locks_remove_file+0x239/0x370
   blkdev_put+0x72/0x2c0
   blkdev_close+0x8d/0xd0
   __fput+0x256/0x770
   ? _raw_read_lock_irq+0x40/0x40
   ____fput+0xe/0x10
   task_work_run+0x10c/0x180
   ? filp_close+0xf7/0x140
   exit_to_usermode_loop+0x151/0x170
   do_syscall_64+0x240/0x2e0
   ? prepare_exit_to_usermode+0xd5/0x190
   entry_SYSCALL_64_after_hwframe+0x44/0xa9
  RIP: 0033:0x7f5a79af05d7
  Code: 00 00 0f 05 48 3d 00 f0 ff ff 77 3f c3 66 0f 1f 44 00 00 53 89 fb 48 83 ec 10 e8 c4 fb ff ff 89 df 89 c2 b8 03 00 00 00 0f 05 <48> 3d 00 f0 ff ff 77 2b 89 d7 89 44 24 0c e8 06 fc ff ff 8b 44 24
  RSP: 002b:00007f5a7799c810 EFLAGS: 00000293 ORIG_RAX: 0000000000000003
  RAX: 0000000000000000 RBX: 0000000000000008 RCX: 00007f5a79af05d7
  RDX: 0000000000000000 RSI: 0000000000000001 RDI: 0000000000000008
  RBP: 00007f5a58000f98 R08: 0000000000000002 R09: 00007f5a7935ee80
  R10: 0000000000000000 R11: 0000000000000293 R12: 000055e432447240
  R13: 0000000000000000 R14: 0000000000000001 R15: 000055e4324a9cf0

  Allocated by task 1236:
   save_stack+0x21/0x80
   __kasan_kmalloc.constprop.6+0xab/0xe0
   kasan_kmalloc+0x9/0x10
   kmem_cache_alloc_trace+0x102/0x210
   nvme_init_identify+0x13c3/0x3820
   nvme_loop_configure_admin_queue+0x4fa/0x5e0
   nvme_loop_create_ctrl+0x469/0xf40
   nvmf_dev_write+0x19a3/0x21ab
   __vfs_write+0x66/0x120
   vfs_write+0x154/0x490
   ksys_write+0x104/0x240
   __x64_sys_write+0x73/0xb0
   do_syscall_64+0xa5/0x2e0
   entry_SYSCALL_64_after_hwframe+0x44/0xa9

  Freed by task 329:
   save_stack+0x21/0x80
   __kasan_slab_free+0x129/0x190
   kasan_slab_free+0xe/0x10
   kfree+0xa7/0x200
   nvme_release_subsystem+0x49/0x60
   device_release+0x72/0x1d0
   kobject_put+0x144/0x410
   put_device+0x13/0x20
   klist_class_dev_put+0x31/0x40
   klist_put+0x8f/0xf0
   klist_del+0xe/0x10
   device_del+0x3a7/0x9a0
   nvme_destroy_subsystem+0xf9/0x150
   nvme_free_ctrl+0x280/0x3a0
   device_release+0x72/0x1d0
   kobject_put+0x144/0x410
   put_device+0x13/0x20
   nvme_free_ns+0xc4/0x100
   nvme_release+0xb3/0xe0
   __blkdev_put+0x549/0x6e0
   blkdev_put+0x72/0x2c0
   blkdev_close+0x8d/0xd0
   __fput+0x256/0x770
   ____fput+0xe/0x10
   task_work_run+0x10c/0x180
   exit_to_usermode_loop+0x151/0x170
   do_syscall_64+0x240/0x2e0
   entry_SYSCALL_64_after_hwframe+0x44/0xa9
Fixes 32fd90c40768 ("nvme: change locking for the per-subsystem controller list")
---
 drivers/nvme/host/core.c | 2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)
drivers/nvme/host/core.c +1 -1
index ab1a2b1ec3637..dfbd5872f4422 100644
@@ -2440,6 +2440,7 @@ static int nvme_init_subsystem(struct nvme_ctrl *ctrl, struct nvme_id_ctrl *id)
 		if (ret) {
 			dev_err(ctrl->device,
 				"failed to register subsystem device.\n");
+			put_device(&subsys->dev);
 			goto out_unlock;
 		}
 		ida_init(&subsys->ns_ida);
@@ -2462,7 +2463,6 @@ static int nvme_init_subsystem(struct nvme_ctrl *ctrl, struct nvme_id_ctrl *id)
 	nvme_put_subsystem(subsys);
 out_unlock:
 	mutex_unlock(&nvme_subsystems_lock);
-	put_device(&subsys->dev);
 	return ret;
 }
 
Signature · 1 line
2.20.1
author maintainer
#15 lore raw

PATCH 14/44 nvme: fix a possible deadlock when passthru commands sent to a multipath device

In reply to Sasha Levin

From: Sagi Grimberg <[email protected]>

[ Upstream commit b9156daeb1601d69007b7e50efcf89d69d72ec1d ]

When the user issues a command with side effects, we will end up freezing the namespace request queue when updating disk info (and the same for the corresponding mpath disk node).

However, we are not freezing the mpath node request queue, which means that mpath I/O can still come in and block on blk_queue_enter (called from nvme_ns_head_make_request -> direct_make_request).

This is a deadlock, because blk_queue_enter will block until the inner namespace request queue is unfroze, but that process is blocked because the namespace revalidation is trying to update the mpath disk info and freeze its request queue (which will never complete because of the I/O that is blocked on blk_queue_enter).

Fix this by freezing all the subsystem nsheads request queues before executing the passthru command. Given that these commands are infrequent we should not worry about this temporary I/O freeze to keep things sane.

Here is the matching hang traces:
--

[ 374.465002] INFO: task systemd-udevd:17994 blocked for more than 122 seconds.
[ 374.472975] Not tainted 5.2.0-rc3-mpdebug+ #42
[ 374.478522] "echo 0 > /proc/sys/kernel/hung_task_timeout_secs" disables this message.
[ 374.487274] systemd-udevd D 0 17994 1 0x00000000
[ 374.493407] Call Trace:
[ 374.496145] __schedule+0x2ef/0x620
[ 374.500047] schedule+0x38/0xa0
[ 374.503569] blk_queue_enter+0x139/0x220
[ 374.507959] ? remove_wait_queue+0x60/0x60
[ 374.512540] direct_make_request+0x60/0x130
[ 374.517219] nvme_ns_head_make_request+0x11d/0x420 [nvme_core]
[ 374.523740] ? generic_make_request_checks+0x307/0x6f0
[ 374.529484] generic_make_request+0x10d/0x2e0
[ 374.534356] submit_bio+0x75/0x140
[ 374.538163] ? guard_bio_eod+0x32/0xe0
[ 374.542361] submit_bh_wbc+0x171/0x1b0
[ 374.546553] block_read_full_page+0x1ed/0x330
[ 374.551426] ? check_disk_change+0x70/0x70
[ 374.556008] ? scan_shadow_nodes+0x30/0x30
[ 374.560588] blkdev_readpage+0x18/0x20
[ 374.564783] do_read_cache_page+0x301/0x860
[ 374.569463] ? blkdev_writepages+0x10/0x10
[ 374.574037] ? prep_new_page+0x88/0x130
[ 374.578329] ? get_page_from_freelist+0xa2f/0x1280
[ 374.583688] ? __alloc_pages_nodemask+0x179/0x320
[ 374.588947] read_cache_page+0x12/0x20
[ 374.593142] read_dev_sector+0x2d/0xd0
[ 374.597337] read_lba+0x104/0x1f0
[ 374.601046] find_valid_gpt+0xfa/0x720
[ 374.605243] ? string_nocheck+0x58/0x70
[ 374.609534] ? find_valid_gpt+0x720/0x720
[ 374.614016] efi_partition+0x89/0x430
[ 374.618113] ? string+0x48/0x60
[ 374.621632] ? snprintf+0x49/0x70
[ 374.625339] ? find_valid_gpt+0x720/0x720
[ 374.629828] check_partition+0x116/0x210
[ 374.634214] rescan_partitions+0xb6/0x360
[ 374.638699] __blkdev_reread_part+0x64/0x70
[ 374.643377] blkdev_reread_part+0x23/0x40
[ 374.647860] blkdev_ioctl+0x48c/0x990
[ 374.651956] block_ioctl+0x41/0x50
[ 374.655766] do_vfs_ioctl+0xa7/0x600
[ 374.659766] ? locks_lock_inode_wait+0xb1/0x150
[ 374.664832] ksys_ioctl+0x67/0x90
[ 374.668539] __x64_sys_ioctl+0x1a/0x20
[ 374.672732] do_syscall_64+0x5a/0x1c0
[ 374.676828] entry_SYSCALL_64_after_hwframe+0x44/0xa9
[ 374.738474] INFO: task nvmeadm:49141 blocked for more than 123 seconds.
[ 374.745871] Not tainted 5.2.0-rc3-mpdebug+ #42
[ 374.751419] "echo 0 > /proc/sys/kernel/hung_task_timeout_secs" disables this message.
[ 374.760170] nvmeadm D 0 49141 36333 0x00004080
[ 374.766301] Call Trace:
[ 374.769038] __schedule+0x2ef/0x620
[ 374.772939] schedule+0x38/0xa0
[ 374.776452] blk_mq_freeze_queue_wait+0x59/0x100
[ 374.781614] ? remove_wait_queue+0x60/0x60
[ 374.786192] blk_mq_freeze_queue+0x1a/0x20
[ 374.790773] nvme_update_disk_info.isra.57+0x5f/0x350 [nvme_core]
[ 374.797582] ? nvme_identify_ns.isra.50+0x71/0xc0 [nvme_core]
[ 374.804006] __nvme_revalidate_disk+0xe5/0x110 [nvme_core]
[ 374.810139] nvme_revalidate_disk+0xa6/0x120 [nvme_core]
[ 374.816078] ? nvme_submit_user_cmd+0x11e/0x320 [nvme_core]
[ 374.822299] nvme_user_cmd+0x264/0x370 [nvme_core]
[ 374.827661] nvme_dev_ioctl+0x112/0x1d0 [nvme_core]
[ 374.833114] do_vfs_ioctl+0xa7/0x600
[ 374.837117] ? __audit_syscall_entry+0xdd/0x130
[ 374.842184] ksys_ioctl+0x67/0x90
[ 374.845891] __x64_sys_ioctl+0x1a/0x20
[ 374.850082] do_syscall_64+0x5a/0x1c0
[ 374.854178] entry_SYSCALL_64_after_hwframe+0x44/0xa9

--

---
 drivers/nvme/host/core.c      |  5 +++++
 drivers/nvme/host/multipath.c | 30 ++++++++++++++++++++++++++++++
 drivers/nvme/host/nvme.h      | 12 ++++++++++++
 3 files changed, 47 insertions(+)
drivers/nvme/host/core.c +5 -0
index dfbd5872f4422..05301b94e2fa0 100644
@@ -1268,6 +1268,9 @@ static u32 nvme_passthru_start(struct nvme_ctrl *ctrl, struct nvme_ns *ns,
 	 */
 	if (effects & (NVME_CMD_EFFECTS_LBCC | NVME_CMD_EFFECTS_CSE_MASK)) {
 		mutex_lock(&ctrl->scan_lock);
+		mutex_lock(&ctrl->subsys->lock);
+		nvme_mpath_start_freeze(ctrl->subsys);
+		nvme_mpath_wait_freeze(ctrl->subsys);
 		nvme_start_freeze(ctrl);
 		nvme_wait_freeze(ctrl);
 	}
@@ -1298,6 +1301,8 @@ static void nvme_passthru_end(struct nvme_ctrl *ctrl, u32 effects)
 		nvme_update_formats(ctrl);
 	if (effects & (NVME_CMD_EFFECTS_LBCC | NVME_CMD_EFFECTS_CSE_MASK)) {
 		nvme_unfreeze(ctrl);
+		nvme_mpath_unfreeze(ctrl->subsys);
+		mutex_unlock(&ctrl->subsys->lock);
 		mutex_unlock(&ctrl->scan_lock);
 	}
 	if (effects & NVME_CMD_EFFECTS_CCC)
drivers/nvme/host/multipath.c +30 -0
index e942b3e840687..dafb9e4aa1237 100644
@@ -12,6 +12,36 @@ module_param(multipath, bool, 0444);
 MODULE_PARM_DESC(multipath,
 	"turn on native support for multiple controllers per subsystem");
 
+void nvme_mpath_unfreeze(struct nvme_subsystem *subsys)
+{
+	struct nvme_ns_head *h;
+
+	lockdep_assert_held(&subsys->lock);
+	list_for_each_entry(h, &subsys->nsheads, entry)
+		if (h->disk)
+			blk_mq_unfreeze_queue(h->disk->queue);
+}
+
+void nvme_mpath_wait_freeze(struct nvme_subsystem *subsys)
+{
+	struct nvme_ns_head *h;
+
+	lockdep_assert_held(&subsys->lock);
+	list_for_each_entry(h, &subsys->nsheads, entry)
+		if (h->disk)
+			blk_mq_freeze_queue_wait(h->disk->queue);
+}
+
+void nvme_mpath_start_freeze(struct nvme_subsystem *subsys)
+{
+	struct nvme_ns_head *h;
+
+	lockdep_assert_held(&subsys->lock);
+	list_for_each_entry(h, &subsys->nsheads, entry)
+		if (h->disk)
+			blk_freeze_queue_start(h->disk->queue);
+}
+
 /*
  * If multipathing is enabled we need to always use the subsystem instance
  * number for numbering our devices to avoid conflicts between subsystems that
drivers/nvme/host/nvme.h +12 -0
index 7391cd0a7739e..b8b45822f7be0 100644
@@ -477,6 +477,9 @@ static inline bool nvme_ctrl_use_ana(struct nvme_ctrl *ctrl)
 	return ctrl->ana_log_buf != NULL;
 }
 
+void nvme_mpath_unfreeze(struct nvme_subsystem *subsys);
+void nvme_mpath_wait_freeze(struct nvme_subsystem *subsys);
+void nvme_mpath_start_freeze(struct nvme_subsystem *subsys);
 void nvme_set_disk_name(char *disk_name, struct nvme_ns *ns,
 			struct nvme_ctrl *ctrl, int *flags);
 void nvme_failover_req(struct request *req);
@@ -555,6 +558,15 @@ static inline void nvme_mpath_uninit(struct nvme_ctrl *ctrl)
 static inline void nvme_mpath_stop(struct nvme_ctrl *ctrl)
 {
 }
+static inline void nvme_mpath_unfreeze(struct nvme_subsystem *subsys)
+{
+}
+static inline void nvme_mpath_wait_freeze(struct nvme_subsystem *subsys)
+{
+}
+static inline void nvme_mpath_start_freeze(struct nvme_subsystem *subsys)
+{
+}
 #endif /* CONFIG_NVME_MULTIPATH */
 
 #ifdef CONFIG_NVM
Signature · 1 line
2.20.1
author maintainer
#16 lore raw

PATCH 15/44 nvme-rdma: fix possible use-after-free in connect error flow

In reply to Sasha Levin

From: Sagi Grimberg <[email protected]>

[ Upstream commit d94211b8bad3787e0655a67284105f57db728cb1 ]

When start_queue fails, we need to make sure to drain the queue cq before freeing the rdma resources because we might still race with the completion path. Have start_queue() error path safely stop the queue.

--

[30371.808111] nvme nvme1: Failed reconnect attempt 11
[30371.808113] nvme nvme1: Reconnecting in 10 seconds...
[...]
[30382.069315] nvme nvme1: creating 4 I/O queues.
[30382.257058] nvme nvme1: Connect Invalid SQE Parameter, qid 4
[30382.257061] nvme nvme1: failed to connect queue: 4 ret=386
[30382.305001] BUG: unable to handle kernel NULL pointer dereference at 0000000000000018
[30382.305022] IP: qedr_poll_cq+0x8a3/0x1170 [qedr]
[30382.305028] PGD 0 P4D 0
[30382.305037] Oops: 0000 [#1] SMP PTI
[...]
[30382.305153] Call Trace:
[30382.305166]  ? __switch_to_asm+0x34/0x70
[30382.305187]  __ib_process_cq+0x56/0xd0 [ib_core]
[30382.305201]  ib_poll_handler+0x26/0x70 [ib_core]
[30382.305213]  irq_poll_softirq+0x88/0x110
[30382.305223]  ? sort_range+0x20/0x20
[30382.305232]  __do_softirq+0xde/0x2c6
[30382.305241]  ? sort_range+0x20/0x20
[30382.305249]  run_ksoftirqd+0x1c/0x60
[30382.305258]  smpboot_thread_fn+0xef/0x160
[30382.305265]  kthread+0x113/0x130
[30382.305273]  ? kthread_create_worker_on_cpu+0x50/0x50
[30382.305281]  ret_from_fork+0x35/0x40

--

---
 drivers/nvme/host/rdma.c | 16 +++++++++++-----
 1 file changed, 11 insertions(+), 5 deletions(-)
drivers/nvme/host/rdma.c +11 -5
index 97f668a39ae1c..7b074323bcdf2 100644
@@ -562,13 +562,17 @@ static int nvme_rdma_alloc_queue(struct nvme_rdma_ctrl *ctrl,
 	return ret;
 }
 
+static void __nvme_rdma_stop_queue(struct nvme_rdma_queue *queue)
+{
+	rdma_disconnect(queue->cm_id);
+	ib_drain_qp(queue->qp);
+}
+
 static void nvme_rdma_stop_queue(struct nvme_rdma_queue *queue)
 {
 	if (!test_and_clear_bit(NVME_RDMA_Q_LIVE, &queue->flags))
 		return;
-
-	rdma_disconnect(queue->cm_id);
-	ib_drain_qp(queue->qp);
+	__nvme_rdma_stop_queue(queue);
 }
 
 static void nvme_rdma_free_queue(struct nvme_rdma_queue *queue)
@@ -607,11 +611,13 @@ static int nvme_rdma_start_queue(struct nvme_rdma_ctrl *ctrl, int idx)
 	else
 		ret = nvmf_connect_admin_queue(&ctrl->ctrl);
 
-	if (!ret)
+	if (!ret) {
 		set_bit(NVME_RDMA_Q_LIVE, &queue->flags);
-	else
+	} else {
+		__nvme_rdma_stop_queue(queue);
 		dev_info(ctrl->ctrl.device,
 			"failed to connect queue: %d ret=%d\n", idx, ret);
+	}
 	return ret;
 }
 
Signature · 1 line
2.20.1
author maintainer
#17 lore raw

PATCH 16/44 nvme: fix controller removal race with scan work

In reply to Sasha Levin

From: Sagi Grimberg <[email protected]>

[ Upstream commit 0157ec8dad3c8fc9bc9790f76e0831ffdaf2e7f0 ]

With multipath enabled, nvme_scan_work() can read from the device (through nvme_mpath_add_disk()) and hang [1]. However, with fabrics, once ctrl->state is set to NVME_CTRL_DELETING, the reads will hang (see nvmf_check_ready()) and the mpath stack device make_request will block if head->list is not empty. However, when the head->list consistst of only DELETING/DEAD controllers, we should actually not block, but rather fail immediately.

In addition, before we go ahead and remove the namespaces, make sure to clear the current path and kick the requeue list so that the request will fast fail upon requeuing.

[1]:
--

  INFO: task kworker/u4:3:166 blocked for more than 120 seconds.
        Not tainted 5.2.0-rc6-vmlocalyes-00005-g808c8c2dc0cf #316
  "echo 0 > /proc/sys/kernel/hung_task_timeout_secs" disables this message.
  kworker/u4:3    D    0   166      2 0x80004000
  Workqueue: nvme-wq nvme_scan_work
  Call Trace:
   __schedule+0x851/0x1400
   schedule+0x99/0x210
   io_schedule+0x21/0x70
   do_read_cache_page+0xa57/0x1330
   read_cache_page+0x4a/0x70
   read_dev_sector+0xbf/0x380
   amiga_partition+0xc4/0x1230
   check_partition+0x30f/0x630
   rescan_partitions+0x19a/0x980
   __blkdev_get+0x85a/0x12f0
   blkdev_get+0x2a5/0x790
   __device_add_disk+0xe25/0x1250
   device_add_disk+0x13/0x20
   nvme_mpath_set_live+0x172/0x2b0
   nvme_update_ns_ana_state+0x130/0x180
   nvme_set_ns_ana_state+0x9a/0xb0
   nvme_parse_ana_log+0x1c3/0x4a0
   nvme_mpath_add_disk+0x157/0x290
   nvme_validate_ns+0x1017/0x1bd0
   nvme_scan_work+0x44d/0x6a0
   process_one_work+0x7d7/0x1240
   worker_thread+0x8e/0xff0
   kthread+0x2c3/0x3b0
   ret_from_fork+0x35/0x40

   INFO: task kworker/u4:1:1034 blocked for more than 120 seconds.
        Not tainted 5.2.0-rc6-vmlocalyes-00005-g808c8c2dc0cf #316
  "echo 0 > /proc/sys/kernel/hung_task_timeout_secs" disables this message.
  kworker/u4:1    D    0  1034      2 0x80004000
  Workqueue: nvme-delete-wq nvme_delete_ctrl_work
  Call Trace:
   __schedule+0x851/0x1400
   schedule+0x99/0x210
   schedule_timeout+0x390/0x830
   wait_for_completion+0x1a7/0x310
   __flush_work+0x241/0x5d0
   flush_work+0x10/0x20
   nvme_remove_namespaces+0x85/0x3d0
   nvme_do_delete_ctrl+0xb4/0x1e0
   nvme_delete_ctrl_work+0x15/0x20
   process_one_work+0x7d7/0x1240
   worker_thread+0x8e/0xff0
   kthread+0x2c3/0x3b0
   ret_from_fork+0x35/0x40

--

---
 drivers/nvme/host/core.c      |  7 ++++++
 drivers/nvme/host/multipath.c | 46 ++++++++++++++++++++++++++++++-----
 drivers/nvme/host/nvme.h      |  9 +++++--
 3 files changed, 54 insertions(+), 8 deletions(-)
drivers/nvme/host/core.c +7 -0
index 05301b94e2fa0..601509b3251ae 100644
@@ -3529,6 +3529,13 @@ void nvme_remove_namespaces(struct nvme_ctrl *ctrl)
 	struct nvme_ns *ns, *next;
 	LIST_HEAD(ns_list);
 
+	/*
+	 * make sure to requeue I/O to all namespaces as these
+	 * might result from the scan itself and must complete
+	 * for the scan_work to make progress
+	 */
+	nvme_mpath_clear_ctrl_paths(ctrl);
+
 	/* prevent racing with ns scanning */
 	flush_work(&ctrl->scan_work);
 
drivers/nvme/host/multipath.c +40 -6

78 lines not on this page · raw message

drivers/nvme/host/nvme.h +7 -2
index b8b45822f7be0..81215ca32671a 100644
@@ -490,7 +490,8 @@ void nvme_mpath_remove_disk(struct nvme_ns_head *head);
 int nvme_mpath_init(struct nvme_ctrl *ctrl, struct nvme_id_ctrl *id);
 void nvme_mpath_uninit(struct nvme_ctrl *ctrl);
 void nvme_mpath_stop(struct nvme_ctrl *ctrl);
-void nvme_mpath_clear_current_path(struct nvme_ns *ns);
+bool nvme_mpath_clear_current_path(struct nvme_ns *ns);
+void nvme_mpath_clear_ctrl_paths(struct nvme_ctrl *ctrl);
 struct nvme_ns *nvme_find_path(struct nvme_ns_head *head);
 
 static inline void nvme_mpath_check_last_path(struct nvme_ns *ns)
@@ -538,7 +539,11 @@ static inline void nvme_mpath_add_disk(struct nvme_ns *ns,
 static inline void nvme_mpath_remove_disk(struct nvme_ns_head *head)
 {
 }
-static inline void nvme_mpath_clear_current_path(struct nvme_ns *ns)
+static inline bool nvme_mpath_clear_current_path(struct nvme_ns *ns)
+{
+	return false;
+}
+static inline void nvme_mpath_clear_ctrl_paths(struct nvme_ctrl *ctrl)
 {
 }
 static inline void nvme_mpath_check_last_path(struct nvme_ns *ns)
Signature · 1 line
2.20.1
author maintainer
#18 lore raw

PATCH 17/44 nvme-pci: Fix async probe remove race

In reply to Sasha Levin

From: Keith Busch <[email protected]>

[ Upstream commit bd46a90634302bfe791e93ad5496f98f165f7ae0 ]

Ensure the controller is not in the NEW state when nvme_probe() exits. This will always allow a subsequent nvme_remove() to set the state to DELETING, fixing a potential race between the initial asynchronous probe and device removal.

---
 drivers/nvme/host/pci.c | 3 ++-
 1 file changed, 2 insertions(+), 1 deletion(-)
drivers/nvme/host/pci.c +2 -1
index f9959eaaa185e..09ffd21d18096 100644
@@ -2712,7 +2712,7 @@ static void nvme_async_probe(void *data, async_cookie_t cookie)
 {
 	struct nvme_dev *dev = data;
 
-	nvme_reset_ctrl_sync(&dev->ctrl);
+	flush_work(&dev->ctrl.reset_work);
 	flush_work(&dev->ctrl.scan_work);
 	nvme_put_ctrl(&dev->ctrl);
 }
@@ -2778,6 +2778,7 @@ static int nvme_probe(struct pci_dev *pdev, const struct pci_device_id *id)
 
 	dev_info(dev->ctrl.device, "pci function %s\n", dev_name(&pdev->dev));
 
+	nvme_reset_ctrl(&dev->ctrl);
 	nvme_get_ctrl(&dev->ctrl);
 	async_schedule(nvme_async_probe, dev);
 
Signature · 1 line
2.20.1
author maintainer
#19 lore raw

PATCH 18/44 soundwire: cadence_master: fix register definition for SLAVE_STATE

In reply to Sasha Levin

From: Pierre-Louis Bossart <[email protected]>

[ Upstream commit b07dd9b400981f487940a4d84292d3a0e7cd9362 ]

wrong prefix and wrong macro.

---
 drivers/soundwire/cadence_master.c | 4 ++--
 1 file changed, 2 insertions(+), 2 deletions(-)
drivers/soundwire/cadence_master.c +2 -2
index 682789bb8ab30..18afb2e21dc9a 100644
@@ -80,8 +80,8 @@
 
 #define CDNS_MCP_INTSET				0x4C
 
-#define CDNS_SDW_SLAVE_STAT			0x50
-#define CDNS_MCP_SLAVE_STAT_MASK		BIT(1, 0)
+#define CDNS_MCP_SLAVE_STAT			0x50
+#define CDNS_MCP_SLAVE_STAT_MASK		GENMASK(1, 0)
 
 #define CDNS_MCP_SLAVE_INTSTAT0			0x54
 #define CDNS_MCP_SLAVE_INTSTAT1			0x58
Signature · 1 line
2.20.1
author maintainer
#20 lore raw

PATCH 19/44 soundwire: cadence_master: fix definitions for INTSTAT0/1

In reply to Sasha Levin

From: Pierre-Louis Bossart <[email protected]>

[ Upstream commit 664b16589f882202b8fa8149d0074f3159bade76 ]

Two off-by-one errors: INTSTAT0 missed BIT(31) and INTSTAT1 is only defined on first 16 bits.

---
 drivers/soundwire/cadence_master.c | 4 ++--
 1 file changed, 2 insertions(+), 2 deletions(-)
drivers/soundwire/cadence_master.c +2 -2
index 18afb2e21dc9a..57ed2e2024bf4 100644
@@ -95,8 +95,8 @@
 #define CDNS_MCP_SLAVE_INTMASK0			0x5C
 #define CDNS_MCP_SLAVE_INTMASK1			0x60
 
-#define CDNS_MCP_SLAVE_INTMASK0_MASK		GENMASK(30, 0)
-#define CDNS_MCP_SLAVE_INTMASK1_MASK		GENMASK(16, 0)
+#define CDNS_MCP_SLAVE_INTMASK0_MASK		GENMASK(31, 0)
+#define CDNS_MCP_SLAVE_INTMASK1_MASK		GENMASK(15, 0)
 
 #define CDNS_MCP_PORT_INTSTAT			0x64
 #define CDNS_MCP_PDI_STAT			0x6C
Signature · 1 line
2.20.1
author maintainer
#21 lore raw

PATCH 20/44 iio: adc: max9611: Fix temperature reading in probe

In reply to Sasha Levin

From: Jacopo Mondi <[email protected]>

[ Upstream commit b9ddd5091160793ee9fac10da765cf3f53d2aaf0 ]

The max9611 driver reads the die temperature at probe time to validate the communication channel. Use the actual read value to perform the test instead of the read function return value, which was mistakenly used so far.

The temperature reading test was only successful because the 0 return value is in the range of supported temperatures.

Fixes 69780a3bbc0b ("iio: adc: Add Maxim max9611 ADC driver")
---
 drivers/iio/adc/max9611.c | 2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)
drivers/iio/adc/max9611.c +1 -1
index 0e3c6529fc4c9..da073d72f649f 100644
@@ -480,7 +480,7 @@ static int max9611_init(struct max9611_dev *max9611)
 	if (ret)
 		return ret;
 
-	regval = ret & MAX9611_TEMP_MASK;
+	regval &= MAX9611_TEMP_MASK;
 
 	if ((regval > MAX9611_TEMP_MAX_POS &&
 	     regval < MAX9611_TEMP_MIN_NEG) ||
Signature · 1 line
2.20.1
author maintainer
#22 lore raw

PATCH 21/44 auxdisplay: panel: need to delete scan_timer when misc_register fails in panel_attach

In reply to Sasha Levin

From: zhengbin <[email protected]>

[ Upstream commit b33d567560c1aadf3033290d74d4fd67af47aa61 ]

In panel_attach, if misc_register fails, we need to delete scan_timer, which was setup in keypad_init->init_scan_timer.

---
 drivers/auxdisplay/panel.c | 2 ++
 1 file changed, 2 insertions(+)
drivers/auxdisplay/panel.c +2 -0
index e06de63497cf8..e6bd727da503a 100644
@@ -1617,6 +1617,8 @@ static void panel_attach(struct parport *port)
 	return;
 
 err_lcd_unreg:
+	if (scan_timer.function)
+		del_timer_sync(&scan_timer);
 	if (lcd.enabled)
 		charlcd_unregister(lcd.charlcd);
 err_unreg_device:
Signature · 1 line
2.20.1
author maintainer
#23 lore raw

PATCH 22/44 btrfs: trim: Check the range passed into to prevent overflow

In reply to Sasha Levin

From: Qu Wenruo <[email protected]>

[ Upstream commit 07301df7d2fc220d3de5f7ad804dcb941400cb00 ]

Normally the range->len is set to default value (U64_MAX), but when it's not default value, we should check if the range overflows.

And if it overflows, return -EINVAL before doing anything.

---
 fs/btrfs/extent-tree.c | 14 +++++++++++---
 1 file changed, 11 insertions(+), 3 deletions(-)
fs/btrfs/extent-tree.c +11 -3
index 5faf057f6f37f..b8f4720879021 100644
@@ -11226,6 +11226,7 @@ int btrfs_trim_fs(struct btrfs_fs_info *fs_info, struct fstrim_range *range)
 	struct btrfs_device *device;
 	struct list_head *devices;
 	u64 group_trimmed;
+	u64 range_end = U64_MAX;
 	u64 start;
 	u64 end;
 	u64 trimmed = 0;
@@ -11235,16 +11236,23 @@ int btrfs_trim_fs(struct btrfs_fs_info *fs_info, struct fstrim_range *range)
 	int dev_ret = 0;
 	int ret = 0;
 
+	/*
+	 * Check range overflow if range->len is set.
+	 * The default range->len is U64_MAX.
+	 */
+	if (range->len != U64_MAX &&
+	    check_add_overflow(range->start, range->len, &range_end))
+		return -EINVAL;
+
 	cache = btrfs_lookup_first_block_group(fs_info, range->start);
 	for (; cache; cache = next_block_group(cache)) {
-		if (cache->key.objectid >= (range->start + range->len)) {
+		if (cache->key.objectid >= range_end) {
 			btrfs_put_block_group(cache);
 			break;
 		}
 
 		start = max(range->start, cache->key.objectid);
-		end = min(range->start + range->len,
-				cache->key.objectid + cache->key.offset);
+		end = min(range_end, cache->key.objectid + cache->key.offset);
 
 		if (end - start >= range->minlen) {
 			if (!block_group_cache_done(cache)) {
Signature · 1 line
2.20.1
author maintainer
#24 lore raw

PATCH 23/44 IB/mlx5: Fix implicit MR release flow

In reply to Sasha Levin

From: Yishai Hadas <[email protected]>

[ Upstream commit f591822c3cf314442819486f45ff7dc1f690e0c0 ]

Once implicit MR is being called to be released by ib_umem_notifier_release() its leaves were marked as "dying".

However, when dereg_mr()->mlx5_ib_free_implicit_mr()->mr_leaf_free() is called, it skips running the mr_leaf_free_action (i.e. umem_odp->work) when those leaves were marked as "dying".

As such ib_umem_release() for the leaves won't be called and their MRs will be leaked as well.

When an application exits/killed without calling dereg_mr we might hit the above flow.

This fatal scenario is reported by WARN_ON() upon mlx5_ib_dealloc_ucontext() as ibcontext->per_mm_list is not empty, the call trace can be seen below.

Originally the "dying" mark as part of ib_umem_notifier_release() was introduced to prevent pagefault_mr() from returning a success response once this happened. However, we already have today the completion mechanism so no need for that in those flows any more. Even in case a success response will be returned the firmware will not find the pages and an error will be returned in the following call as a released mm will cause ib_umem_odp_map_dma_pages() to permanently fail mmget_not_zero().

Fix the above issue by dropping the "dying" from the above flows. The other flows that are using "dying" are still needed it for their synchronization purposes.

   WARNING: CPU: 1 PID: 7218 at
   drivers/infiniband/hw/mlx5/main.c:2004
		  mlx5_ib_dealloc_ucontext+0x84/0x90 [mlx5_ib]
   CPU: 1 PID: 7218 Comm: ibv_rc_pingpong Tainted: G     E
	       5.2.0-rc6+ #13
   Call Trace:
   uverbs_destroy_ufile_hw+0xb5/0x120 [ib_uverbs]
   ib_uverbs_close+0x1f/0x80 [ib_uverbs]
   __fput+0xbe/0x250
   task_work_run+0x88/0xa0
   do_exit+0x2cb/0xc30
   ? __fput+0x14b/0x250
   do_group_exit+0x39/0xb0
   get_signal+0x191/0x920
   ? _raw_spin_unlock_bh+0xa/0x20
   ? inet_csk_accept+0x229/0x2f0
   do_signal+0x36/0x5e0
   ? put_unused_fd+0x5b/0x70
   ? __sys_accept4+0x1a6/0x1e0
   ? inet_hash+0x35/0x40
   ? release_sock+0x43/0x90
   ? _raw_spin_unlock_bh+0xa/0x20
   ? inet_listen+0x9f/0x120
   exit_to_usermode_loop+0x5c/0xc6
   do_syscall_64+0x182/0x1b0
   entry_SYSCALL_64_after_hwframe+0x44/0xa9
---
 drivers/infiniband/core/umem_odp.c |  4 ----
 drivers/infiniband/hw/mlx5/odp.c   | 24 +++++++++---------------
 2 files changed, 9 insertions(+), 19 deletions(-)
drivers/infiniband/core/umem_odp.c +0 -4
index e4b13a32692a9..5e5f7dd82c50d 100644
@@ -114,10 +114,6 @@ static int ib_umem_notifier_release_trampoline(struct ib_umem_odp *umem_odp,
 	 * prevent any further fault handling on this MR.
 	 */
 	ib_umem_notifier_start_account(umem_odp);
-	umem_odp->dying = 1;
-	/* Make sure that the fact the umem is dying is out before we release
-	 * all pending page faults. */
-	smp_wmb();
 	complete_all(&umem_odp->notifier_completion);
 	umem->context->invalidate_range(umem_odp, ib_umem_start(umem),
 					ib_umem_end(umem));
drivers/infiniband/hw/mlx5/odp.c +9 -15
index 91507a2e92900..d711f3e31fa74 100644
@@ -581,7 +581,6 @@ static int pagefault_mr(struct mlx5_ib_dev *dev, struct mlx5_ib_mr *mr,
 			u32 flags)
 {
 	int npages = 0, current_seq, page_shift, ret, np;
-	bool implicit = false;
 	struct ib_umem_odp *odp_mr = to_ib_umem_odp(mr->umem);
 	bool downgrade = flags & MLX5_PF_FLAGS_DOWNGRADE;
 	bool prefetch = flags & MLX5_PF_FLAGS_PREFETCH;
@@ -596,7 +595,6 @@ static int pagefault_mr(struct mlx5_ib_dev *dev, struct mlx5_ib_mr *mr,
 		if (IS_ERR(odp))
 			return PTR_ERR(odp);
 		mr = odp->private;
-		implicit = true;
 	} else {
 		odp = odp_mr;
 	}
@@ -684,19 +682,15 @@ static int pagefault_mr(struct mlx5_ib_dev *dev, struct mlx5_ib_mr *mr,
 
 out:
 	if (ret == -EAGAIN) {
-		if (implicit || !odp->dying) {
-			unsigned long timeout =
-				msecs_to_jiffies(MMU_NOTIFIER_TIMEOUT);
-
-			if (!wait_for_completion_timeout(
-					&odp->notifier_completion,
-					timeout)) {
-				mlx5_ib_warn(dev, "timeout waiting for mmu notifier. seq %d against %d. notifiers_count=%d\n",
-					     current_seq, odp->notifiers_seq, odp->notifiers_count);
-			}
-		} else {
-			/* The MR is being killed, kill the QP as well. */
-			ret = -EFAULT;
+		unsigned long timeout = msecs_to_jiffies(MMU_NOTIFIER_TIMEOUT);
+
+		if (!wait_for_completion_timeout(&odp->notifier_completion,
+						 timeout)) {
+			mlx5_ib_warn(
+				dev,
+				"timeout waiting for mmu notifier. seq %d against %d. notifiers_count=%d\n",
+				current_seq, odp->notifiers_seq,
+				odp->notifiers_count);
 		}
 	}
 
Signature · 1 line
2.20.1
author maintainer
#25 lore raw

PATCH 24/44 dmaengine: stm32-mdma: Fix a possible null-pointer dereference in stm32_mdma_irq_handler()

In reply to Sasha Levin

From: Jia-Ju Bai <[email protected]>

[ Upstream commit 39c71a5b8212f4b502d9a630c6706ac723abd422 ]

In stm32_mdma_irq_handler(), chan is checked on line 1368. When chan is NULL, it is still used on line 1369:

    dev_err(chan2dev(chan), "MDMA channel not initialized\n");

Thus, a possible null-pointer dereference may occur.

To fix this bug, "dev_dbg(mdma2dev(dmadev), ...)" is used instead.

Fixes a4ffb13c8946 ("dmaengine: Add STM32 MDMA driver")
---
 drivers/dma/stm32-mdma.c | 2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)
drivers/dma/stm32-mdma.c +1 -1
index d6e919d3936a2..1311de74bfdde 100644
@@ -1366,7 +1366,7 @@ static irqreturn_t stm32_mdma_irq_handler(int irq, void *devid)
 
 	chan = &dmadev->chan[id];
 	if (!chan) {
-		dev_err(chan2dev(chan), "MDMA channel not initialized\n");
+		dev_dbg(mdma2dev(dmadev), "MDMA channel not initialized\n");
 		goto exit;
 	}
 
Signature · 1 line
2.20.1
author maintainer
#26 lore raw

PATCH 25/44 omap-dma/omap_vout_vrfb: fix off-by-one fi value

In reply to Sasha Levin

From: Hans Verkuil <[email protected]>

[ Upstream commit d555c34338cae844b207564c482e5a3fb089d25e ]

The OMAP 4 TRM specifies that when using double-index addressing the address increases by the ES plus the EI value minus 1 within a frame. When a full frame is transferred, the address increases by the ES plus the frame index (FI) value minus 1.

The omap-dma code didn't account for the 'minus 1' in the FI register. To get correct addressing, add 1 to the src_icg value.

This was found when testing a hacked version of the media m2m-deinterlace.c driver on a Pandaboard.

The only other source that uses this feature is omap_vout_vrfb.c, and that adds a + 1 when setting the dst_icg. This is a workaround for the broken omap-dma.c behavior. So remove the workaround at the same time that we fix omap-dma.c.

I tested the omap_vout driver with a Beagle XM board to check that the '+ 1' in omap_vout_vrfb.c was indeed a workaround for the omap-dma bug.

---
 drivers/dma/ti/omap-dma.c                    | 4 ++--
 drivers/media/platform/omap/omap_vout_vrfb.c | 3 +--
 2 files changed, 3 insertions(+), 4 deletions(-)
drivers/dma/ti/omap-dma.c +2 -2
index ba2489d4ea246..ba27802efcd0a 100644
@@ -1234,7 +1234,7 @@ static struct dma_async_tx_descriptor *omap_dma_prep_dma_interleaved(
 	if (src_icg) {
 		d->ccr |= CCR_SRC_AMODE_DBLIDX;
 		d->ei = 1;
-		d->fi = src_icg;
+		d->fi = src_icg + 1;
 	} else if (xt->src_inc) {
 		d->ccr |= CCR_SRC_AMODE_POSTINC;
 		d->fi = 0;
@@ -1249,7 +1249,7 @@ static struct dma_async_tx_descriptor *omap_dma_prep_dma_interleaved(
 	if (dst_icg) {
 		d->ccr |= CCR_DST_AMODE_DBLIDX;
 		sg->ei = 1;
-		sg->fi = dst_icg;
+		sg->fi = dst_icg + 1;
 	} else if (xt->dst_inc) {
 		d->ccr |= CCR_DST_AMODE_POSTINC;
 		sg->fi = 0;
drivers/media/platform/omap/omap_vout_vrfb.c +1 -2
index 29e3f5da59c1f..11ec048929e80 100644
@@ -253,8 +253,7 @@ int omap_vout_prepare_vrfb(struct omap_vout_device *vout,
 	 */
 
 	pixsize = vout->bpp * vout->vrfb_bpp;
-	dst_icg = ((MAX_PIXELS_PER_LINE * pixsize) -
-		  (vout->pix.width * vout->bpp)) + 1;
+	dst_icg = MAX_PIXELS_PER_LINE * pixsize - vout->pix.width * vout->bpp;
 
 	xt->src_start = vout->buf_phy_addr[vb->i];
 	xt->dst_start = vout->vrfb_context[vb->i].paddr[0];
Signature · 1 line
2.20.1
author maintainer
#27 lore raw

PATCH 26/44 iommu/dma: Handle SG length overflow better

In reply to Sasha Levin

From: Robin Murphy <[email protected]>

[ Upstream commit ab2cbeb0ed301a9f0460078e91b09f39958212ef ]

Since scatterlist dimensions are all unsigned ints, in the relatively rare cases where a device's max_segment_size is set to UINT_MAX, then the "cur_len + s_length <= max_len" check in __finalise_sg() will always return true. As a result, the corner case of such a device mapping an excessively large scatterlist which is mergeable to or beyond a total length of 4GB can lead to overflow and a bogus truncated dma_length in the resulting segment.

As we already assume that any single segment must be no longer than max_len to begin with, this can easily be addressed by reshuffling the comparison.

Fixes 809eac54cdd6 ("iommu/dma: Implement scatterlist segment merging")
---
 drivers/iommu/dma-iommu.c | 2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)
drivers/iommu/dma-iommu.c +1 -1
index 379318266468c..8c02d2283d647 100644
@@ -710,7 +710,7 @@ static int __finalise_sg(struct device *dev, struct scatterlist *sg, int nents,
 		 * - and wouldn't make the resulting output segment too long
 		 */
 		if (cur_len && !s_iova_off && (dma_addr & seg_mask) &&
-		    (cur_len + s_length <= max_len)) {
+		    (max_len - cur_len >= s_length)) {
 			/* ...then concatenate it with the previous one */
 			cur_len += s_length;
 		} else {
Signature · 6 lines
2.20.1

_______________________________________________
iommu mailing list
[email protected]
https://lists.linuxfoundation.org/mailman/listinfo/iommu
author maintainer
#28 lore raw

PATCH 27/44 dma-direct: don't truncate dma_required_mask to bus addressing capabilities

In reply to Sasha Levin

From: Lucas Stach <[email protected]>

[ Upstream commit d8ad55538abe443919e20e0bb996561bca9cad84 ]

The dma required_mask needs to reflect the actual addressing capabilities needed to handle the whole system RAM. When truncated down to the bus addressing capabilities dma_addressing_limited() will incorrectly signal no limitations for devices which are restricted by the bus_dma_mask.

Fixes b4ebe6063204 (dma-direct: implement complete bus_dma_mask handling)
---
 kernel/dma/direct.c | 3 ---
 1 file changed, 3 deletions(-)
kernel/dma/direct.c +0 -3
index 2c2772e9702ab..9912be7a970de 100644
@@ -55,9 +55,6 @@ u64 dma_direct_get_required_mask(struct device *dev)
 {
 	u64 max_dma = phys_to_dma_direct(dev, (max_pfn - 1) << PAGE_SHIFT);
 
-	if (dev->bus_dma_mask && dev->bus_dma_mask < max_dma)
-		max_dma = dev->bus_dma_mask;
-
 	return (1ULL << (fls64(max_dma) - 1)) * 2 - 1;
 }
 
Signature · 6 lines
2.20.1

_______________________________________________
iommu mailing list
[email protected]
https://lists.linuxfoundation.org/mailman/listinfo/iommu
author maintainer
#29 lore raw

PATCH 28/44 usb: gadget: composite: Clear "suspended" on reset/disconnect

In reply to Sasha Levin

From: Benjamin Herrenschmidt <[email protected]>

[ Upstream commit 602fda17c7356bb7ae98467d93549057481d11dd ]

In some cases, one can get out of suspend with a reset or a disconnect followed by a reconnect. Previously we would leave a stale suspended flag set.

---
 drivers/usb/gadget/composite.c | 1 +
 1 file changed, 1 insertion(+)
drivers/usb/gadget/composite.c +1 -0
index b8a15840b4ffd..dfcabadeed01b 100644
@@ -1976,6 +1976,7 @@ void composite_disconnect(struct usb_gadget *gadget)
 	 * disconnect callbacks?
 	 */
 	spin_lock_irqsave(&cdev->lock, flags);
+	cdev->suspended = 0;
 	if (cdev->config)
 		reset_config(cdev);
 	if (cdev->driver->disconnect)
Signature · 1 line
2.20.1
author maintainer
#30 lore raw

PATCH 29/44 usb: gadget: mass_storage: Fix races between fsg_disable and fsg_set_alt

In reply to Sasha Levin

From: Benjamin Herrenschmidt <[email protected]>

[ Upstream commit 4a56a478a525d6427be90753451c40e1327caa1a ]

If fsg_disable() and fsg_set_alt() are called too closely to each other (for example due to a quick reset/reconnect), what can happen is that fsg_set_alt sets common->new_fsg from an interrupt while handle_exception is trying to process the config change caused by fsg_disable():

fsg_disable()
...
handle_exception()
sets state back to FSG_STATE_NORMAL hasn't yet called do_set_interface() or is inside it.

---> interrupt

	fsg_set_alt
		sets common->new_fsg
		queues a new FSG_STATE_CONFIG_CHANGE

<---

Now, the first handle_exception can "see" the updated new_fsg, treats it as if it was a fsg_set_alt() response, call usb_composite_setup_continue() etc...

But then, the thread sees the second FSG_STATE_CONFIG_CHANGE, and goes back down the same path, wipes and reattaches a now active fsg, and .. calls usb_composite_setup_continue() which at this point is wrong.

Not only we get a backtrace, but I suspect the second set_interface wrecks some state causing the host to get upset in my case.

This fixes it by replacing "new_fsg" by a "state argument" (same principle) which is set in the same lock section as the state update, and retrieved similarly.

That way, there is never any discrepancy between the dequeued state and the observed value of it. We keep the ability to have the latest reconfig operation take precedence, but we guarantee that once "dequeued" the argument (new_fsg) will not be clobbered by any new event.

---
 drivers/usb/gadget/function/f_mass_storage.c | 28 +++++++++++++-------
 1 file changed, 18 insertions(+), 10 deletions(-)
drivers/usb/gadget/function/f_mass_storage.c +18 -10
index 043f97ad8f226..f2bc8d0370676 100644
@@ -261,7 +261,7 @@ struct fsg_common;
 struct fsg_common {
 	struct usb_gadget	*gadget;
 	struct usb_composite_dev *cdev;
-	struct fsg_dev		*fsg, *new_fsg;
+	struct fsg_dev		*fsg;
 	wait_queue_head_t	io_wait;
 	wait_queue_head_t	fsg_wait;
 
@@ -290,6 +290,7 @@ struct fsg_common {
 	unsigned int		bulk_out_maxpacket;
 	enum fsg_state		state;		/* For exception handling */
 	unsigned int		exception_req_tag;
+	void			*exception_arg;
 
 	enum data_direction	data_dir;
 	u32			data_size;
@@ -391,7 +392,8 @@ static int fsg_set_halt(struct fsg_dev *fsg, struct usb_ep *ep)
 
 /* These routines may be called in process context or in_irq */
 
-static void raise_exception(struct fsg_common *common, enum fsg_state new_state)
+static void __raise_exception(struct fsg_common *common, enum fsg_state new_state,
+			      void *arg)
 {
 	unsigned long		flags;
 
@@ -404,6 +406,7 @@ static void raise_exception(struct fsg_common *common, enum fsg_state new_state)
 	if (common->state <= new_state) {
 		common->exception_req_tag = common->ep0_req_tag;
 		common->state = new_state;
+		common->exception_arg = arg;
 		if (common->thread_task)
 			send_sig_info(SIGUSR1, SEND_SIG_PRIV,
 				      common->thread_task);
@@ -411,6 +414,10 @@ static void raise_exception(struct fsg_common *common, enum fsg_state new_state)
 	spin_unlock_irqrestore(&common->lock, flags);
 }
 
+static void raise_exception(struct fsg_common *common, enum fsg_state new_state)
+{
+	__raise_exception(common, new_state, NULL);
+}
 
 /*-------------------------------------------------------------------------*/
 
@@ -2285,16 +2292,16 @@ static int do_set_interface(struct fsg_common *common, struct fsg_dev *new_fsg)
 static int fsg_set_alt(struct usb_function *f, unsigned intf, unsigned alt)
 {
 	struct fsg_dev *fsg = fsg_from_func(f);
-	fsg->common->new_fsg = fsg;
-	raise_exception(fsg->common, FSG_STATE_CONFIG_CHANGE);
+
+	__raise_exception(fsg->common, FSG_STATE_CONFIG_CHANGE, fsg);
 	return USB_GADGET_DELAYED_STATUS;
 }
 
 static void fsg_disable(struct usb_function *f)
 {
 	struct fsg_dev *fsg = fsg_from_func(f);
-	fsg->common->new_fsg = NULL;
-	raise_exception(fsg->common, FSG_STATE_CONFIG_CHANGE);
+
+	__raise_exception(fsg->common, FSG_STATE_CONFIG_CHANGE, NULL);
 }
 
 
@@ -2307,6 +2314,7 @@ static void handle_exception(struct fsg_common *common)
 	enum fsg_state		old_state;
 	struct fsg_lun		*curlun;
 	unsigned int		exception_req_tag;
+	struct fsg_dev		*new_fsg;
 
 	/*
 	 * Clear the existing signals.  Anything but SIGUSR1 is converted
@@ -2360,6 +2368,7 @@ static void handle_exception(struct fsg_common *common)
 	common->next_buffhd_to_fill = &common->buffhds[0];
 	common->next_buffhd_to_drain = &common->buffhds[0];
 	exception_req_tag = common->exception_req_tag;
+	new_fsg = common->exception_arg;
 	old_state = common->state;
 	common->state = FSG_STATE_NORMAL;
 
@@ -2413,8 +2422,8 @@ static void handle_exception(struct fsg_common *common)
 		break;
 
 	case FSG_STATE_CONFIG_CHANGE:
-		do_set_interface(common, common->new_fsg);
-		if (common->new_fsg)
+		do_set_interface(common, new_fsg);
+		if (new_fsg)
 			usb_composite_setup_continue(common->cdev);
 		break;
 
@@ -2989,8 +2998,7 @@ static void fsg_unbind(struct usb_configuration *c, struct usb_function *f)
 
 	DBG(fsg, "unbind\n");
 	if (fsg->common->fsg == fsg) {
-		fsg->common->new_fsg = NULL;
-		raise_exception(fsg->common, FSG_STATE_CONFIG_CHANGE);
+		__raise_exception(fsg->common, FSG_STATE_CONFIG_CHANGE, NULL);
 		/* FIXME: make interruptible or killable somehow? */
 		wait_event(common->fsg_wait, common->fsg != fsg);
 	}
Signature · 1 line
2.20.1
author maintainer
#31 lore raw

PATCH 30/44 habanalabs: fix DRAM usage accounting on context tear down

In reply to Sasha Levin

From: Tomer Tayar <[email protected]>

[ Upstream commit c8113756ba27298d6e95403c087dc5881b419a99 ]

The patch fix the DRAM usage accounting by adding a missing update of the DRAM memory consumption, when a context is being torn down without an organized release of the allocated memory.

---
 drivers/misc/habanalabs/memory.c | 2 ++
 1 file changed, 2 insertions(+)
drivers/misc/habanalabs/memory.c +2 -0
index 693877e37fd87..924a438ba9736 100644
@@ -1629,6 +1629,8 @@ void hl_vm_ctx_fini(struct hl_ctx *ctx)
 			dev_dbg(hdev->dev,
 				"page list 0x%p of asid %d is still alive\n",
 				phys_pg_list, ctx->asid);
+			atomic64_sub(phys_pg_list->total_size,
+					&hdev->dram_used_mem);
 			free_phys_pg_pack(hdev, phys_pg_list);
 			idr_remove(&vm->phys_pg_pack_handles, i);
 		}
Signature · 1 line
2.20.1
author maintainer
#32 lore raw

PATCH 31/44 habanalabs: fix endianness handling for packets from user

In reply to Sasha Levin

From: Ben Segal <[email protected]>

[ Upstream commit 213ad5ad016a0da975b35f54e8cd236c3b04724b ]

Packets that arrive from the user and need to be parsed by the driver are assumed to be in LE format.

This patch fix all the places where the code handles these packets and use the correct endianness macros to handle them, as the driver handles the packets in CPU format (LE or BE depending on the arch).

---
 drivers/misc/habanalabs/goya/goya.c           | 32 +++++++++++--------
 .../habanalabs/include/goya/goya_packets.h    | 13 ++++++++
 2 files changed, 32 insertions(+), 13 deletions(-)
drivers/misc/habanalabs/goya/goya.c +19 -13
index 02d116b01a1a2..0644fd7742057 100644
@@ -3425,12 +3425,13 @@ static int goya_validate_cb(struct hl_device *hdev,
 	while (cb_parsed_length < parser->user_cb_size) {
 		enum packet_id pkt_id;
 		u16 pkt_size;
-		void *user_pkt;
+		struct goya_packet *user_pkt;
 
-		user_pkt = (void *) (uintptr_t)
+		user_pkt = (struct goya_packet *) (uintptr_t)
 			(parser->user_cb->kernel_address + cb_parsed_length);
 
-		pkt_id = (enum packet_id) (((*(u64 *) user_pkt) &
+		pkt_id = (enum packet_id) (
+				(le64_to_cpu(user_pkt->header) &
 				PACKET_HEADER_PACKET_ID_MASK) >>
 					PACKET_HEADER_PACKET_ID_SHIFT);
 
@@ -3450,7 +3451,8 @@ static int goya_validate_cb(struct hl_device *hdev,
 			 * need to validate here as well because patch_cb() is
 			 * not called in MMU path while this function is called
 			 */
-			rc = goya_validate_wreg32(hdev, parser, user_pkt);
+			rc = goya_validate_wreg32(hdev,
+				parser, (struct packet_wreg32 *) user_pkt);
 			break;
 
 		case PACKET_WREG_BULK:
@@ -3478,10 +3480,10 @@ static int goya_validate_cb(struct hl_device *hdev,
 		case PACKET_LIN_DMA:
 			if (is_mmu)
 				rc = goya_validate_dma_pkt_mmu(hdev, parser,
-						user_pkt);
+					(struct packet_lin_dma *) user_pkt);
 			else
 				rc = goya_validate_dma_pkt_no_mmu(hdev, parser,
-						user_pkt);
+					(struct packet_lin_dma *) user_pkt);
 			break;
 
 		case PACKET_MSG_LONG:
@@ -3654,15 +3656,16 @@ static int goya_patch_cb(struct hl_device *hdev,
 		enum packet_id pkt_id;
 		u16 pkt_size;
 		u32 new_pkt_size = 0;
-		void *user_pkt, *kernel_pkt;
+		struct goya_packet *user_pkt, *kernel_pkt;
 
-		user_pkt = (void *) (uintptr_t)
+		user_pkt = (struct goya_packet *) (uintptr_t)
 			(parser->user_cb->kernel_address + cb_parsed_length);
-		kernel_pkt = (void *) (uintptr_t)
+		kernel_pkt = (struct goya_packet *) (uintptr_t)
 			(parser->patched_cb->kernel_address +
 					cb_patched_cur_length);
 
-		pkt_id = (enum packet_id) (((*(u64 *) user_pkt) &
+		pkt_id = (enum packet_id) (
+				(le64_to_cpu(user_pkt->header) &
 				PACKET_HEADER_PACKET_ID_MASK) >>
 					PACKET_HEADER_PACKET_ID_SHIFT);
 
@@ -3677,15 +3680,18 @@ static int goya_patch_cb(struct hl_device *hdev,
 
 		switch (pkt_id) {
 		case PACKET_LIN_DMA:
-			rc = goya_patch_dma_packet(hdev, parser, user_pkt,
-						kernel_pkt, &new_pkt_size);
+			rc = goya_patch_dma_packet(hdev, parser,
+					(struct packet_lin_dma *) user_pkt,
+					(struct packet_lin_dma *) kernel_pkt,
+					&new_pkt_size);
 			cb_patched_cur_length += new_pkt_size;
 			break;
 
 		case PACKET_WREG_32:
 			memcpy(kernel_pkt, user_pkt, pkt_size);
 			cb_patched_cur_length += pkt_size;
-			rc = goya_validate_wreg32(hdev, parser, kernel_pkt);
+			rc = goya_validate_wreg32(hdev, parser,
+					(struct packet_wreg32 *) kernel_pkt);
 			break;
 
 		case PACKET_WREG_BULK:
drivers/misc/habanalabs/include/goya/goya_packets.h +13 -0
index a14407b975e4e..ef54bad205099 100644
@@ -52,6 +52,19 @@ enum goya_dma_direction {
 #define GOYA_PKT_CTL_MB_SHIFT		31
 #define GOYA_PKT_CTL_MB_MASK		0x80000000
 
+/* All packets have, at least, an 8-byte header, which contains
+ * the packet type. The kernel driver uses the packet header for packet
+ * validation and to perform any necessary required preparation before
+ * sending them off to the hardware.
+ */
+struct goya_packet {
+	__le64 header;
+	/* The rest of the packet data follows. Use the corresponding
+	 * packet_XXX struct to deference the data, based on packet type
+	 */
+	u8 contents[0];
+};
+
 struct packet_nop {
 	__le32 reserved;
 	__le32 ctl;
Signature · 1 line
2.20.1
author maintainer
#33 lore raw

PATCH 32/44 habanalabs: fix completion queue handling when host is BE

In reply to Sasha Levin

From: Ben Segal <[email protected]>

[ Upstream commit 4e87334a0ef43663019dbaf3638ad10fd8c3320c ]

This patch fix the CQ irq handler to work in hosts with BE architecture. It adds the correct endian-swapping macros around the relevant memory accesses.

---
 drivers/misc/habanalabs/irq.c | 27 +++++++++++++--------------
 1 file changed, 13 insertions(+), 14 deletions(-)
drivers/misc/habanalabs/irq.c +13 -14
index ea9f72ff456cf..199791b57caf2 100644
@@ -80,8 +80,7 @@ irqreturn_t hl_irq_handler_cq(int irq, void *arg)
 	struct hl_cs_job *job;
 	bool shadow_index_valid;
 	u16 shadow_index;
-	u32 *cq_entry;
-	u32 *cq_base;
+	struct hl_cq_entry *cq_entry, *cq_base;
 
 	if (hdev->disabled) {
 		dev_dbg(hdev->dev,
@@ -90,29 +89,29 @@ irqreturn_t hl_irq_handler_cq(int irq, void *arg)
 		return IRQ_HANDLED;
 	}
 
-	cq_base = (u32 *) (uintptr_t) cq->kernel_address;
+	cq_base = (struct hl_cq_entry *) (uintptr_t) cq->kernel_address;
 
 	while (1) {
-		bool entry_ready = ((cq_base[cq->ci] & CQ_ENTRY_READY_MASK)
+		bool entry_ready = ((le32_to_cpu(cq_base[cq->ci].data) &
+					CQ_ENTRY_READY_MASK)
 						>> CQ_ENTRY_READY_SHIFT);
 
 		if (!entry_ready)
 			break;
 
-		cq_entry = (u32 *) &cq_base[cq->ci];
+		cq_entry = (struct hl_cq_entry *) &cq_base[cq->ci];
 
-		/*
-		 * Make sure we read CQ entry contents after we've
+		/* Make sure we read CQ entry contents after we've
 		 * checked the ownership bit.
 		 */
 		dma_rmb();
 
-		shadow_index_valid =
-			((*cq_entry & CQ_ENTRY_SHADOW_INDEX_VALID_MASK)
+		shadow_index_valid = ((le32_to_cpu(cq_entry->data) &
+					CQ_ENTRY_SHADOW_INDEX_VALID_MASK)
 					>> CQ_ENTRY_SHADOW_INDEX_VALID_SHIFT);
 
-		shadow_index = (u16)
-			((*cq_entry & CQ_ENTRY_SHADOW_INDEX_MASK)
+		shadow_index = (u16) ((le32_to_cpu(cq_entry->data) &
+					CQ_ENTRY_SHADOW_INDEX_MASK)
 					>> CQ_ENTRY_SHADOW_INDEX_SHIFT);
 
 		queue = &hdev->kernel_queues[cq->hw_queue_id];
@@ -122,8 +121,7 @@ irqreturn_t hl_irq_handler_cq(int irq, void *arg)
 			queue_work(hdev->cq_wq, &job->finish_work);
 		}
 
-		/*
-		 * Update ci of the context's queue. There is no
+		/* Update ci of the context's queue. There is no
 		 * need to protect it with spinlock because this update is
 		 * done only inside IRQ and there is a different IRQ per
 		 * queue
@@ -131,7 +129,8 @@ irqreturn_t hl_irq_handler_cq(int irq, void *arg)
 		queue->ci = hl_queue_inc_ptr(queue->ci);
 
 		/* Clear CQ entry ready bit */
-		cq_base[cq->ci] &= ~CQ_ENTRY_READY_MASK;
+		cq_entry->data = cpu_to_le32(le32_to_cpu(cq_entry->data) &
+						~CQ_ENTRY_READY_MASK);
 
 		cq->ci = hl_cq_inc_ptr(cq->ci);
 
Signature · 1 line
2.20.1
author maintainer
#34 lore raw

PATCH 33/44 habanalabs: fix endianness handling for internal QMAN submission

In reply to Sasha Levin

From: Oded Gabbay <[email protected]>

[ Upstream commit b9040c99414ba5b85090595a61abc686a5dbb388 ]

The PQs of internal H/W queues (QMANs) can be located in different memory areas for different ASICs. Therefore, when writing PQEs, we need to use the correct function according to the location of the PQ. e.g. if the PQ is located in the device's memory (SRAM or DRAM), we need to use memcpy_toio() so it would work in architectures that have separate address ranges for IO memory.

This patch makes the code that writes the PQE to be ASIC-specific so we can handle this properly per ASIC.

---
 drivers/misc/habanalabs/goya/goya.c  |  7 ++++---
 drivers/misc/habanalabs/goya/goyaP.h |  2 +-
 drivers/misc/habanalabs/habanalabs.h |  9 +++++++--
 drivers/misc/habanalabs/hw_queue.c   | 14 +++++---------
 4 files changed, 17 insertions(+), 15 deletions(-)
drivers/misc/habanalabs/goya/goya.c +4 -3
index 0644fd7742057..9216cc3599178 100644
@@ -2716,9 +2716,10 @@ void goya_ring_doorbell(struct hl_device *hdev, u32 hw_queue_id, u32 pi)
 				GOYA_ASYNC_EVENT_ID_PI_UPDATE);
 }
 
-void goya_flush_pq_write(struct hl_device *hdev, u64 *pq, u64 exp_val)
+void goya_pqe_write(struct hl_device *hdev, __le64 *pqe, struct hl_bd *bd)
 {
-	/* Not needed in Goya */
+	/* The QMANs are on the SRAM so need to copy to IO space */
+	memcpy_toio((void __iomem *) pqe, bd, sizeof(struct hl_bd));
 }
 
 static void *goya_dma_alloc_coherent(struct hl_device *hdev, size_t size,
@@ -4784,7 +4785,7 @@ static const struct hl_asic_funcs goya_funcs = {
 	.resume = goya_resume,
 	.cb_mmap = goya_cb_mmap,
 	.ring_doorbell = goya_ring_doorbell,
-	.flush_pq_write = goya_flush_pq_write,
+	.pqe_write = goya_pqe_write,
 	.asic_dma_alloc_coherent = goya_dma_alloc_coherent,
 	.asic_dma_free_coherent = goya_dma_free_coherent,
 	.get_int_queue_base = goya_get_int_queue_base,
drivers/misc/habanalabs/goya/goyaP.h +1 -1
index c83cab0d641e2..e2040fd331ca1 100644
@@ -170,7 +170,7 @@ int goya_late_init(struct hl_device *hdev);
 void goya_late_fini(struct hl_device *hdev);
 
 void goya_ring_doorbell(struct hl_device *hdev, u32 hw_queue_id, u32 pi);
-void goya_flush_pq_write(struct hl_device *hdev, u64 *pq, u64 exp_val);
+void goya_pqe_write(struct hl_device *hdev, __le64 *pqe, struct hl_bd *bd);
 void goya_update_eq_ci(struct hl_device *hdev, u32 val);
 void goya_restore_phase_topology(struct hl_device *hdev);
 int goya_context_switch(struct hl_device *hdev, u32 asid);
drivers/misc/habanalabs/habanalabs.h +7 -2
index adef7d9d7488a..d56ab65d5b2a4 100644
@@ -449,7 +449,11 @@ enum hl_pll_frequency {
  * @resume: handles IP specific H/W or SW changes for resume.
  * @cb_mmap: maps a CB.
  * @ring_doorbell: increment PI on a given QMAN.
- * @flush_pq_write: flush PQ entry write if necessary, WARN if flushing failed.
+ * @pqe_write: Write the PQ entry to the PQ. This is ASIC-specific
+ *             function because the PQs are located in different memory areas
+ *             per ASIC (SRAM, DRAM, Host memory) and therefore, the method of
+ *             writing the PQE must match the destination memory area
+ *             properties.
  * @asic_dma_alloc_coherent: Allocate coherent DMA memory by calling
  *                           dma_alloc_coherent(). This is ASIC function because
  *                           its implementation is not trivial when the driver
@@ -518,7 +522,8 @@ struct hl_asic_funcs {
 	int (*cb_mmap)(struct hl_device *hdev, struct vm_area_struct *vma,
 			u64 kaddress, phys_addr_t paddress, u32 size);
 	void (*ring_doorbell)(struct hl_device *hdev, u32 hw_queue_id, u32 pi);
-	void (*flush_pq_write)(struct hl_device *hdev, u64 *pq, u64 exp_val);
+	void (*pqe_write)(struct hl_device *hdev, __le64 *pqe,
+			struct hl_bd *bd);
 	void* (*asic_dma_alloc_coherent)(struct hl_device *hdev, size_t size,
 					dma_addr_t *dma_handle, gfp_t flag);
 	void (*asic_dma_free_coherent)(struct hl_device *hdev, size_t size,
drivers/misc/habanalabs/hw_queue.c +5 -9
index 2894d89759334..bb76794747279 100644
@@ -290,23 +290,19 @@ static void int_hw_queue_schedule_job(struct hl_cs_job *job)
 	struct hl_device *hdev = job->cs->ctx->hdev;
 	struct hl_hw_queue *q = &hdev->kernel_queues[job->hw_queue_id];
 	struct hl_bd bd;
-	u64 *pi, *pbd = (u64 *) &bd;
+	__le64 *pi;
 
 	bd.ctl = 0;
-	bd.len = __cpu_to_le32(job->job_cb_size);
-	bd.ptr = __cpu_to_le64((u64) (uintptr_t) job->user_cb);
+	bd.len = cpu_to_le32(job->job_cb_size);
+	bd.ptr = cpu_to_le64((u64) (uintptr_t) job->user_cb);
 
-	pi = (u64 *) (uintptr_t) (q->kernel_address +
+	pi = (__le64 *) (uintptr_t) (q->kernel_address +
 		((q->pi & (q->int_queue_len - 1)) * sizeof(bd)));
 
-	pi[0] = pbd[0];
-	pi[1] = pbd[1];
-
 	q->pi++;
 	q->pi &= ((q->int_queue_len << 1) - 1);
 
-	/* Flush PQ entry write. Relevant only for specific ASICs */
-	hdev->asic_funcs->flush_pq_write(hdev, pi, pbd[0]);
+	hdev->asic_funcs->pqe_write(hdev, pi, &bd);
 
 	hdev->asic_funcs->ring_doorbell(hdev, q->hw_queue_id, q->pi);
 }
Signature · 1 line
2.20.1
author maintainer
#35 lore raw

PATCH 34/44 habanalabs: fix device IRQ unmasking for BE host

In reply to Sasha Levin

From: Ben Segal <[email protected]>

[ Upstream commit b421d83a3947369fd5718824aecfaebe1efbf7ed ]

When unmasking IRQs inside the ASIC, the driver passes an array of all the IRQ to unmask. The ASIC's CPU is working in LE so when running in a BE host, the driver needs to do the proper endianness swapping when preparing this array.

In addition, this patch also fixes the endianness of a couple of kernel log debug messages that print values of packets

---
 drivers/misc/habanalabs/goya/goya.c | 33 +++++++++++++++++++++--------
 1 file changed, 24 insertions(+), 9 deletions(-)
drivers/misc/habanalabs/goya/goya.c +24 -9
index 9216cc3599178..ac6b252a1ddcb 100644
@@ -3311,9 +3311,11 @@ static int goya_validate_dma_pkt_no_mmu(struct hl_device *hdev,
 	int rc;
 
 	dev_dbg(hdev->dev, "DMA packet details:\n");
-	dev_dbg(hdev->dev, "source == 0x%llx\n", user_dma_pkt->src_addr);
-	dev_dbg(hdev->dev, "destination == 0x%llx\n", user_dma_pkt->dst_addr);
-	dev_dbg(hdev->dev, "size == %u\n", user_dma_pkt->tsize);
+	dev_dbg(hdev->dev, "source == 0x%llx\n",
+		le64_to_cpu(user_dma_pkt->src_addr));
+	dev_dbg(hdev->dev, "destination == 0x%llx\n",
+		le64_to_cpu(user_dma_pkt->dst_addr));
+	dev_dbg(hdev->dev, "size == %u\n", le32_to_cpu(user_dma_pkt->tsize));
 
 	ctl = le32_to_cpu(user_dma_pkt->ctl);
 	user_dir = (ctl & GOYA_PKT_LIN_DMA_CTL_DMA_DIR_MASK) >>
@@ -3342,9 +3344,11 @@ static int goya_validate_dma_pkt_mmu(struct hl_device *hdev,
 				struct packet_lin_dma *user_dma_pkt)
 {
 	dev_dbg(hdev->dev, "DMA packet details:\n");
-	dev_dbg(hdev->dev, "source == 0x%llx\n", user_dma_pkt->src_addr);
-	dev_dbg(hdev->dev, "destination == 0x%llx\n", user_dma_pkt->dst_addr);
-	dev_dbg(hdev->dev, "size == %u\n", user_dma_pkt->tsize);
+	dev_dbg(hdev->dev, "source == 0x%llx\n",
+		le64_to_cpu(user_dma_pkt->src_addr));
+	dev_dbg(hdev->dev, "destination == 0x%llx\n",
+		le64_to_cpu(user_dma_pkt->dst_addr));
+	dev_dbg(hdev->dev, "size == %u\n", le32_to_cpu(user_dma_pkt->tsize));
 
 	/*
 	 * WA for HW-23.
@@ -3384,7 +3388,8 @@ static int goya_validate_wreg32(struct hl_device *hdev,
 
 	dev_dbg(hdev->dev, "WREG32 packet details:\n");
 	dev_dbg(hdev->dev, "reg_offset == 0x%x\n", reg_offset);
-	dev_dbg(hdev->dev, "value      == 0x%x\n", wreg_pkt->value);
+	dev_dbg(hdev->dev, "value      == 0x%x\n",
+		le32_to_cpu(wreg_pkt->value));
 
 	if (reg_offset != (mmDMA_CH_0_WR_COMP_ADDR_LO & 0x1FFF)) {
 		dev_err(hdev->dev, "WREG32 packet with illegal address 0x%x\n",
@@ -4252,6 +4257,8 @@ static int goya_unmask_irq_arr(struct hl_device *hdev, u32 *irq_arr,
 	size_t total_pkt_size;
 	long result;
 	int rc;
+	int irq_num_entries, irq_arr_index;
+	__le32 *goya_irq_arr;
 
 	total_pkt_size = sizeof(struct armcp_unmask_irq_arr_packet) +
 			irq_arr_size;
@@ -4269,8 +4276,16 @@ static int goya_unmask_irq_arr(struct hl_device *hdev, u32 *irq_arr,
 	if (!pkt)
 		return -ENOMEM;
 
-	pkt->length = cpu_to_le32(irq_arr_size / sizeof(irq_arr[0]));
-	memcpy(&pkt->irqs, irq_arr, irq_arr_size);
+	irq_num_entries = irq_arr_size / sizeof(irq_arr[0]);
+	pkt->length = cpu_to_le32(irq_num_entries);
+
+	/* We must perform any necessary endianness conversation on the irq
+	 * array being passed to the goya hardware
+	 */
+	for (irq_arr_index = 0, goya_irq_arr = (__le32 *) &pkt->irqs;
+			irq_arr_index < irq_num_entries ; irq_arr_index++)
+		goya_irq_arr[irq_arr_index] =
+				cpu_to_le32(irq_arr[irq_arr_index]);
 
 	pkt->armcp_pkt.ctl = cpu_to_le32(ARMCP_PACKET_UNMASK_RAZWI_IRQ_ARRAY <<
 						ARMCP_PKT_CTL_OPCODE_SHIFT);
Signature · 1 line
2.20.1
author maintainer
#36 lore raw

PATCH 35/44 xen/blkback: fix memory leaks

In reply to Sasha Levin

From: Wenwen Wang <[email protected]>

[ Upstream commit ae78ca3cf3d9e9f914bfcd0bc5c389ff18b9c2e0 ]

In read_per_ring_refs(), after 'req' and related memory regions are allocated, xen_blkif_map() is invoked to map the shared frame, irq, and etc. However, if this mapping process fails, no cleanup is performed, leading to memory leaks. To fix this issue, invoke the cleanup before returning the error.

---
 drivers/block/xen-blkback/xenbus.c | 6 +++---
 1 file changed, 3 insertions(+), 3 deletions(-)
drivers/block/xen-blkback/xenbus.c +3 -3
index 3ac6a5d180717..b90dbcd99c03e 100644
@@ -965,6 +965,7 @@ static int read_per_ring_refs(struct xen_blkif_ring *ring, const char *dir)
 		}
 	}
 
+	err = -ENOMEM;
 	for (i = 0; i < nr_grefs * XEN_BLKIF_REQS_PER_PAGE; i++) {
 		req = kzalloc(sizeof(*req), GFP_KERNEL);
 		if (!req)
@@ -987,7 +988,7 @@ static int read_per_ring_refs(struct xen_blkif_ring *ring, const char *dir)
 	err = xen_blkif_map(ring, ring_ref, nr_grefs, evtchn);
 	if (err) {
 		xenbus_dev_fatal(dev, err, "mapping ring-ref port %u", evtchn);
-		return err;
+		goto fail;
 	}
 
 	return 0;
@@ -1007,8 +1008,7 @@ static int read_per_ring_refs(struct xen_blkif_ring *ring, const char *dir)
 		}
 		kfree(req);
 	}
-	return -ENOMEM;
-
+	return err;
 }
 
 static int connect_ring(struct backend_info *be)
Signature · 1 line
2.20.1
author maintainer
#37 lore raw

PATCH 36/44 arm64: cpufeature: Don't treat granule sizes as strict

In reply to Sasha Levin

From: Will Deacon <[email protected]>

[ Upstream commit 5717fe5ab38f9ccb32718bcb03bea68409c9cce4 ]

If a CPU doesn't support the page size for which the kernel is configured, then we will complain and refuse to bring it online. For secondary CPUs (and the boot CPU on a system booting with EFI), we will also print an error identifying the mismatch.

Consequently, the only time that the cpufeature code can detect a granule size mismatch is for a granule other than the one that is currently being used. Although we would rather such systems didn't exist, we've unfortunately lost that battle and Kevin reports that on his amlogic S922X (odroid-n2 board) we end up warning and taining with defconfig because 16k pages are not supported by all of the CPUs.

In such a situation, we don't actually care about the feature mismatch, particularly now that KVM only exposes the sanitised view of the CPU registers (commit 93390c0a1b20 - "arm64: KVM: Hide unsupported AArch64 CPU features from guests"). Treat the granule fields as non-strict and let Kevin run without a tainted kernel.

---
 arch/arm64/kernel/cpufeature.c | 14 +++++++++++---
 1 file changed, 11 insertions(+), 3 deletions(-)
arch/arm64/kernel/cpufeature.c +11 -3
index ae63eedea1c12..68faf535f40a3 100644
@@ -184,9 +184,17 @@ static const struct arm64_ftr_bits ftr_id_aa64zfr0[] = {
 };
 
 static const struct arm64_ftr_bits ftr_id_aa64mmfr0[] = {
-	S_ARM64_FTR_BITS(FTR_HIDDEN, FTR_STRICT, FTR_LOWER_SAFE, ID_AA64MMFR0_TGRAN4_SHIFT, 4, ID_AA64MMFR0_TGRAN4_NI),
-	S_ARM64_FTR_BITS(FTR_HIDDEN, FTR_STRICT, FTR_LOWER_SAFE, ID_AA64MMFR0_TGRAN64_SHIFT, 4, ID_AA64MMFR0_TGRAN64_NI),
-	ARM64_FTR_BITS(FTR_HIDDEN, FTR_STRICT, FTR_LOWER_SAFE, ID_AA64MMFR0_TGRAN16_SHIFT, 4, ID_AA64MMFR0_TGRAN16_NI),
+	/*
+	 * We already refuse to boot CPUs that don't support our configured
+	 * page size, so we can only detect mismatches for a page size other
+	 * than the one we're currently using. Unfortunately, SoCs like this
+	 * exist in the wild so, even though we don't like it, we'll have to go
+	 * along with it and treat them as non-strict.
+	 */
+	S_ARM64_FTR_BITS(FTR_HIDDEN, FTR_NONSTRICT, FTR_LOWER_SAFE, ID_AA64MMFR0_TGRAN4_SHIFT, 4, ID_AA64MMFR0_TGRAN4_NI),
+	S_ARM64_FTR_BITS(FTR_HIDDEN, FTR_NONSTRICT, FTR_LOWER_SAFE, ID_AA64MMFR0_TGRAN64_SHIFT, 4, ID_AA64MMFR0_TGRAN64_NI),
+	ARM64_FTR_BITS(FTR_HIDDEN, FTR_NONSTRICT, FTR_LOWER_SAFE, ID_AA64MMFR0_TGRAN16_SHIFT, 4, ID_AA64MMFR0_TGRAN16_NI),
+
 	ARM64_FTR_BITS(FTR_HIDDEN, FTR_STRICT, FTR_LOWER_SAFE, ID_AA64MMFR0_BIGENDEL0_SHIFT, 4, 0),
 	/* Linux shouldn't care about secure memory */
 	ARM64_FTR_BITS(FTR_HIDDEN, FTR_NONSTRICT, FTR_LOWER_SAFE, ID_AA64MMFR0_SNSMEM_SHIFT, 4, 0),
Signature · 1 line
2.20.1
author maintainer
#38 lore raw

PATCH 37/44 riscv: fix flush_tlb_range() end address for flush_tlb_page()

In reply to Sasha Levin

From: Paul Walmsley <[email protected]>

[ Upstream commit eb93685847a9055283d05951c1b205e737f38533 ]

The RISC-V kernel implementation of flush_tlb_page() when CONFIG_SMP is set is wrong. It passes zero to flush_tlb_range() as the final address to flush, but it should be at least 'addr'.

Some other Linux architecture ports use the beginning address to flush, plus PAGE_SIZE, as the final address to flush. This might flush slightly more than what's needed, but it seems unlikely that being more clever would improve anything. So let's just take that implementation for now.

While here, convert the macro into a static inline function, primarily to avoid unintentional multiple evaluations of 'addr'.

This second version of the patch fixes a coding style issue found by Christoph Hellwig <[email protected]>.

---
 arch/riscv/include/asm/tlbflush.h | 11 +++++++++--
 1 file changed, 9 insertions(+), 2 deletions(-)
arch/riscv/include/asm/tlbflush.h +9 -2
index 687dd19735a7e..4d9bbe8438bf6 100644
@@ -53,10 +53,17 @@ static inline void remote_sfence_vma(struct cpumask *cmask, unsigned long start,
 }
 
 #define flush_tlb_all() sbi_remote_sfence_vma(NULL, 0, -1)
-#define flush_tlb_page(vma, addr) flush_tlb_range(vma, addr, 0)
+
 #define flush_tlb_range(vma, start, end) \
 	remote_sfence_vma(mm_cpumask((vma)->vm_mm), start, (end) - (start))
-#define flush_tlb_mm(mm) \
+
+static inline void flush_tlb_page(struct vm_area_struct *vma,
+				  unsigned long addr)
+{
+	flush_tlb_range(vma, addr, addr + PAGE_SIZE);
+}
+
+#define flush_tlb_mm(mm)				\
 	remote_sfence_vma(mm_cpumask(mm), 0, -1)
 
 #endif /* CONFIG_SMP */
Signature · 7 lines
2.20.1


_______________________________________________
linux-riscv mailing list
[email protected]
http://lists.infradead.org/mailman/listinfo/linux-riscv
author maintainer
#39 lore raw

PATCH 38/44 i2c: rcar: avoid race when unregistering slave client

In reply to Sasha Levin

From: Wolfram Sang <[email protected]>

[ Upstream commit 7b814d852af6944657c2961039f404c4490771c0 ]

After we disabled interrupts, there might still be an active one running. Sync before clearing the pointer to the slave device.

Fixes de20d1857dd6 ("i2c: rcar: add slave support")
---
 drivers/i2c/busses/i2c-rcar.c | 11 +++++++----
 1 file changed, 7 insertions(+), 4 deletions(-)
drivers/i2c/busses/i2c-rcar.c +7 -4
index d39a4606f72d3..531c01100b560 100644
@@ -139,6 +139,7 @@ struct rcar_i2c_priv {
 	enum dma_data_direction dma_direction;
 
 	struct reset_control *rstc;
+	int irq;
 };
 
 #define rcar_i2c_priv_to_dev(p)		((p)->adap.dev.parent)
@@ -861,9 +862,11 @@ static int rcar_unreg_slave(struct i2c_client *slave)
 
 	WARN_ON(!priv->slave);
 
+	/* disable irqs and ensure none is running before clearing ptr */
 	rcar_i2c_write(priv, ICSIER, 0);
 	rcar_i2c_write(priv, ICSCR, 0);
 
+	synchronize_irq(priv->irq);
 	priv->slave = NULL;
 
 	pm_runtime_put(rcar_i2c_priv_to_dev(priv));
@@ -918,7 +921,7 @@ static int rcar_i2c_probe(struct platform_device *pdev)
 	struct i2c_adapter *adap;
 	struct device *dev = &pdev->dev;
 	struct i2c_timings i2c_t;
-	int irq, ret;
+	int ret;
 
 	/* Otherwise logic will break because some bytes must always use PIO */
 	BUILD_BUG_ON_MSG(RCAR_MIN_DMA_LEN < 3, "Invalid min DMA length");
@@ -984,10 +987,10 @@ static int rcar_i2c_probe(struct platform_device *pdev)
 		pm_runtime_put(dev);
 
 
-	irq = platform_get_irq(pdev, 0);
-	ret = devm_request_irq(dev, irq, rcar_i2c_irq, 0, dev_name(dev), priv);
+	priv->irq = platform_get_irq(pdev, 0);
+	ret = devm_request_irq(dev, priv->irq, rcar_i2c_irq, 0, dev_name(dev), priv);
 	if (ret < 0) {
-		dev_err(dev, "cannot get irq %d\n", irq);
+		dev_err(dev, "cannot get irq %d\n", priv->irq);
 		goto out_pm_disable;
 	}
 
Signature · 1 line
2.20.1
author maintainer
#40 lore raw

PATCH 39/44 i2c: emev2: avoid race when unregistering slave client

In reply to Sasha Levin

From: Wolfram Sang <[email protected]>

[ Upstream commit d7437fc0d8291181debe032671a289b6bd93f46f ]

After we disabled interrupts, there might still be an active one running. Sync before clearing the pointer to the slave device.

Fixes c31d0a00021d ("i2c: emev2: add slave support")
---
 drivers/i2c/busses/i2c-emev2.c | 16 ++++++++++++----
 1 file changed, 12 insertions(+), 4 deletions(-)
drivers/i2c/busses/i2c-emev2.c +12 -4
index 35b302d983e0d..959d4912ec0d5 100644
@@ -69,6 +69,7 @@ struct em_i2c_device {
 	struct completion msg_done;
 	struct clk *sclk;
 	struct i2c_client *slave;
+	int irq;
 };
 
 static inline void em_clear_set_bit(struct em_i2c_device *priv, u8 clear, u8 set, u8 reg)
@@ -339,6 +340,12 @@ static int em_i2c_unreg_slave(struct i2c_client *slave)
 
 	writeb(0, priv->base + I2C_OFS_SVA0);
 
+	/*
+	 * Wait for interrupt to finish. New slave irqs cannot happen because we
+	 * cleared the slave address and, thus, only extension codes will be
+	 * detected which do not use the slave ptr.
+	 */
+	synchronize_irq(priv->irq);
 	priv->slave = NULL;
 
 	return 0;
@@ -355,7 +362,7 @@ static int em_i2c_probe(struct platform_device *pdev)
 {
 	struct em_i2c_device *priv;
 	struct resource *r;
-	int irq, ret;
+	int ret;
 
 	priv = devm_kzalloc(&pdev->dev, sizeof(*priv), GFP_KERNEL);
 	if (!priv)
@@ -390,8 +397,8 @@ static int em_i2c_probe(struct platform_device *pdev)
 
 	em_i2c_reset(&priv->adap);
 
-	irq = platform_get_irq(pdev, 0);
-	ret = devm_request_irq(&pdev->dev, irq, em_i2c_irq_handler, 0,
+	priv->irq = platform_get_irq(pdev, 0);
+	ret = devm_request_irq(&pdev->dev, priv->irq, em_i2c_irq_handler, 0,
 				"em_i2c", priv);
 	if (ret)
 		goto err_clk;
@@ -401,7 +408,8 @@ static int em_i2c_probe(struct platform_device *pdev)
 	if (ret)
 		goto err_clk;
 
-	dev_info(&pdev->dev, "Added i2c controller %d, irq %d\n", priv->adap.nr, irq);
+	dev_info(&pdev->dev, "Added i2c controller %d, irq %d\n", priv->adap.nr,
+		 priv->irq);
 
 	return 0;
 
Signature · 1 line
2.20.1
author maintainer
#41 lore raw

PATCH 40/44 drm/scheduler: use job count instead of peek

In reply to Sasha Levin

From: Christian König <[email protected]>

[ Upstream commit e1b4ce25dbc93ab0cb8ed0f236a3b9ff7b03802c ]

The spsc_queue_peek function is accessing queue->head which belongs to the consumer thread and shouldn't be accessed by the producer

This is fixing a rare race condition when destroying entities.

---
 drivers/gpu/drm/scheduler/sched_entity.c | 4 ++--
 1 file changed, 2 insertions(+), 2 deletions(-)
drivers/gpu/drm/scheduler/sched_entity.c +2 -2
index 35ddbec1375ae..671c90f34ede6 100644
@@ -95,7 +95,7 @@ static bool drm_sched_entity_is_idle(struct drm_sched_entity *entity)
 	rmb(); /* for list_empty to work without lock */
 
 	if (list_empty(&entity->list) ||
-	    spsc_queue_peek(&entity->job_queue) == NULL)
+	    spsc_queue_count(&entity->job_queue) == 0)
 		return true;
 
 	return false;
@@ -281,7 +281,7 @@ void drm_sched_entity_fini(struct drm_sched_entity *entity)
 	/* Consumption of existing IBs wasn't completed. Forcefully
 	 * remove them here.
 	 */
-	if (spsc_queue_peek(&entity->job_queue)) {
+	if (spsc_queue_count(&entity->job_queue)) {
 		if (sched) {
 			/* Park the kernel for a moment to make sure it isn't processing
 			 * our enity.
Signature · 1 line
2.20.1
author maintainer
#42 lore raw

PATCH 41/44 drm/ast: Fixed reboot test may cause system hanged

In reply to Sasha Levin

From: "Y.C. Chen" <[email protected]>

[ Upstream commit 05b439711f6ff8700e8660f97a1179650778b9cb ]

There is another thread still access standard VGA I/O while loading drm driver. Disable standard VGA I/O decode to avoid this issue.

---
 drivers/gpu/drm/ast/ast_main.c | 5 ++++-
 drivers/gpu/drm/ast/ast_mode.c | 2 +-
 drivers/gpu/drm/ast/ast_post.c | 2 +-
 3 files changed, 6 insertions(+), 3 deletions(-)
drivers/gpu/drm/ast/ast_main.c +4 -1
index 2854399856ba0..4aebe21e6ad9f 100644
@@ -131,8 +131,8 @@ static int ast_detect_chip(struct drm_device *dev, bool *need_post)
 
 
 	/* Enable extended register access */
-	ast_enable_mmio(dev);
 	ast_open_key(ast);
+	ast_enable_mmio(dev);
 
 	/* Find out whether P2A works or whether to use device-tree */
 	ast_detect_config_mode(dev, &scu_rev);
@@ -576,6 +576,9 @@ void ast_driver_unload(struct drm_device *dev)
 {
 	struct ast_private *ast = dev->dev_private;
 
+	/* enable standard VGA decode */
+	ast_set_index_reg(ast, AST_IO_CRTC_PORT, 0xa1, 0x04);
+
 	ast_release_firmware(dev);
 	kfree(ast->dp501_fw_addr);
 	ast_mode_fini(dev);
drivers/gpu/drm/ast/ast_mode.c +1 -1
index 97fed0627d1c8..74da15a3341a8 100644
@@ -601,7 +601,7 @@ static int ast_crtc_mode_set(struct drm_crtc *crtc,
 		return -EINVAL;
 	ast_open_key(ast);
 
-	ast_set_index_reg_mask(ast, AST_IO_CRTC_PORT, 0xa1, 0xff, 0x04);
+	ast_set_index_reg(ast, AST_IO_CRTC_PORT, 0xa1, 0x06);
 
 	ast_set_std_reg(crtc, adjusted_mode, &vbios_mode);
 	ast_set_crtc_reg(crtc, adjusted_mode, &vbios_mode);
drivers/gpu/drm/ast/ast_post.c +1 -1
index f7d421359d564..c1d1ac51d1c20 100644
@@ -46,7 +46,7 @@ void ast_enable_mmio(struct drm_device *dev)
 {
 	struct ast_private *ast = dev->dev_private;
 
-	ast_set_index_reg_mask(ast, AST_IO_CRTC_PORT, 0xa1, 0xff, 0x04);
+	ast_set_index_reg(ast, AST_IO_CRTC_PORT, 0xa1, 0x06);
 }
 
 
Signature · 6 lines
2.20.1

_______________________________________________
dri-devel mailing list
[email protected]
https://lists.freedesktop.org/mailman/listinfo/dri-devel
author maintainer
#43 lore raw

PATCH 42/44 usb: host: fotg2: restart hcd after port reset

In reply to Sasha Levin

From: Hans Ulli Kroll <[email protected]>

[ Upstream commit 777758888ffe59ef754cc39ab2f275dc277732f4 ]

On the Gemini SoC the FOTG2 stalls after port reset so restart the HCD after each port reset.

---
 drivers/usb/host/fotg210-hcd.c | 4 ++++
 1 file changed, 4 insertions(+)
drivers/usb/host/fotg210-hcd.c +4 -0
index 0da68df259c86..7bf621d40c5ae 100644
@@ -1628,6 +1628,10 @@ static int fotg210_hub_control(struct usb_hcd *hcd, u16 typeReq, u16 wValue,
 			/* see what we found out */
 			temp = check_reset_complete(fotg210, wIndex, status_reg,
 					fotg210_readl(fotg210, status_reg));
+
+			/* restart schedule */
+			fotg210->command |= CMD_RUN;
+			fotg210_writel(fotg210, fotg210->command, &fotg210->regs->command);
 		}
 
 		if (!(temp & (PORT_RESUME|PORT_RESET))) {
Signature · 1 line
2.20.1
author maintainer
#44 lore raw

PATCH 43/44 tools: hv: fixed Python pep8/flake8 warnings for lsvmbus

In reply to Sasha Levin

From: Adrian Vladu <[email protected]>

[ Upstream commit 5912e791f3018de0a007c8cfa9cb38c97d3e5f5c ]

Fixed pep8/flake8 python style code for lsvmbus tool.

The TAB indentation was on purpose ignored (pep8 rule W191) to make sure the code is complying with the Linux code guideline. The following command doe not show any warnings now:
pep8 --ignore=W191 lsvmbus
flake8 --ignore=W191 lsvmbus

Cc "K. Y. Srinivasan" <[email protected]>
Cc Haiyang Zhang <[email protected]>
Cc Stephen Hemminger <[email protected]>
Cc Sasha Levin <[email protected]>
Cc Dexuan Cui <[email protected]>
Cc Alessandro Pilotti <[email protected]>
---
 tools/hv/lsvmbus | 75 +++++++++++++++++++++++++++---------------------
 1 file changed, 42 insertions(+), 33 deletions(-)
tools/hv/lsvmbus +42 -33
index 55e7374bade0d..099f2c44dbed2 100644
@@ -4,10 +4,10 @@
 import os
 from optparse import OptionParser
 
+help_msg = "print verbose messages. Try -vv, -vvv for  more verbose messages"
 parser = OptionParser()
-parser.add_option("-v", "--verbose", dest="verbose",
-		   help="print verbose messages. Try -vv, -vvv for \
-			more verbose messages", action="count")
+parser.add_option(
+	"-v", "--verbose", dest="verbose", help=help_msg, action="count")
 
 (options, args) = parser.parse_args()
 
@@ -21,27 +21,28 @@ if not os.path.isdir(vmbus_sys_path):
 	exit(-1)
 
 vmbus_dev_dict = {
-	'{0e0b6031-5213-4934-818b-38d90ced39db}' : '[Operating system shutdown]',
-	'{9527e630-d0ae-497b-adce-e80ab0175caf}' : '[Time Synchronization]',
-	'{57164f39-9115-4e78-ab55-382f3bd5422d}' : '[Heartbeat]',
-	'{a9a0f4e7-5a45-4d96-b827-8a841e8c03e6}' : '[Data Exchange]',
-	'{35fa2e29-ea23-4236-96ae-3a6ebacba440}' : '[Backup (volume checkpoint)]',
-	'{34d14be3-dee4-41c8-9ae7-6b174977c192}' : '[Guest services]',
-	'{525074dc-8985-46e2-8057-a307dc18a502}' : '[Dynamic Memory]',
-	'{cfa8b69e-5b4a-4cc0-b98b-8ba1a1f3f95a}' : 'Synthetic mouse',
-	'{f912ad6d-2b17-48ea-bd65-f927a61c7684}' : 'Synthetic keyboard',
-	'{da0a7802-e377-4aac-8e77-0558eb1073f8}' : 'Synthetic framebuffer adapter',
-	'{f8615163-df3e-46c5-913f-f2d2f965ed0e}' : 'Synthetic network adapter',
-	'{32412632-86cb-44a2-9b5c-50d1417354f5}' : 'Synthetic IDE Controller',
-	'{ba6163d9-04a1-4d29-b605-72e2ffb1dc7f}' : 'Synthetic SCSI Controller',
-	'{2f9bcc4a-0069-4af3-b76b-6fd0be528cda}' : 'Synthetic fiber channel adapter',
-	'{8c2eaf3d-32a7-4b09-ab99-bd1f1c86b501}' : 'Synthetic RDMA adapter',
-	'{44c4f61d-4444-4400-9d52-802e27ede19f}' : 'PCI Express pass-through',
-	'{276aacf4-ac15-426c-98dd-7521ad3f01fe}' : '[Reserved system device]',
-	'{f8e65716-3cb3-4a06-9a60-1889c5cccab5}' : '[Reserved system device]',
-	'{3375baf4-9e15-4b30-b765-67acb10d607b}' : '[Reserved system device]',
+	'{0e0b6031-5213-4934-818b-38d90ced39db}': '[Operating system shutdown]',
+	'{9527e630-d0ae-497b-adce-e80ab0175caf}': '[Time Synchronization]',
+	'{57164f39-9115-4e78-ab55-382f3bd5422d}': '[Heartbeat]',
+	'{a9a0f4e7-5a45-4d96-b827-8a841e8c03e6}': '[Data Exchange]',
+	'{35fa2e29-ea23-4236-96ae-3a6ebacba440}': '[Backup (volume checkpoint)]',
+	'{34d14be3-dee4-41c8-9ae7-6b174977c192}': '[Guest services]',
+	'{525074dc-8985-46e2-8057-a307dc18a502}': '[Dynamic Memory]',
+	'{cfa8b69e-5b4a-4cc0-b98b-8ba1a1f3f95a}': 'Synthetic mouse',
+	'{f912ad6d-2b17-48ea-bd65-f927a61c7684}': 'Synthetic keyboard',
+	'{da0a7802-e377-4aac-8e77-0558eb1073f8}': 'Synthetic framebuffer adapter',
+	'{f8615163-df3e-46c5-913f-f2d2f965ed0e}': 'Synthetic network adapter',
+	'{32412632-86cb-44a2-9b5c-50d1417354f5}': 'Synthetic IDE Controller',
+	'{ba6163d9-04a1-4d29-b605-72e2ffb1dc7f}': 'Synthetic SCSI Controller',
+	'{2f9bcc4a-0069-4af3-b76b-6fd0be528cda}': 'Synthetic fiber channel adapter',
+	'{8c2eaf3d-32a7-4b09-ab99-bd1f1c86b501}': 'Synthetic RDMA adapter',
+	'{44c4f61d-4444-4400-9d52-802e27ede19f}': 'PCI Express pass-through',
+	'{276aacf4-ac15-426c-98dd-7521ad3f01fe}': '[Reserved system device]',
+	'{f8e65716-3cb3-4a06-9a60-1889c5cccab5}': '[Reserved system device]',
+	'{3375baf4-9e15-4b30-b765-67acb10d607b}': '[Reserved system device]',
 }
 
+
 def get_vmbus_dev_attr(dev_name, attr):
 	try:
 		f = open('%s/%s/%s' % (vmbus_sys_path, dev_name, attr), 'r')
@@ -52,6 +53,7 @@ def get_vmbus_dev_attr(dev_name, attr):
 
 	return lines
 
+
 class VMBus_Dev:
 	pass
 
@@ -66,12 +68,13 @@ for f in os.listdir(vmbus_sys_path):
 
 	chn_vp_mapping = get_vmbus_dev_attr(f, 'channel_vp_mapping')
 	chn_vp_mapping = [c.strip() for c in chn_vp_mapping]
-	chn_vp_mapping = sorted(chn_vp_mapping,
-		key = lambda c : int(c.split(':')[0]))
+	chn_vp_mapping = sorted(
+		chn_vp_mapping, key=lambda c: int(c.split(':')[0]))
 
-	chn_vp_mapping = ['\tRel_ID=%s, target_cpu=%s' %
-				(c.split(':')[0], c.split(':')[1])
-					for c in chn_vp_mapping]
+	chn_vp_mapping = [
+		'\tRel_ID=%s, target_cpu=%s' %
+		(c.split(':')[0], c.split(':')[1]) for c in chn_vp_mapping
+	]
 	d = VMBus_Dev()
 	d.sysfs_path = '%s/%s' % (vmbus_sys_path, f)
 	d.vmbus_id = vmbus_id
@@ -85,7 +88,7 @@ for f in os.listdir(vmbus_sys_path):
 	vmbus_dev_list.append(d)
 
 
-vmbus_dev_list  = sorted(vmbus_dev_list, key = lambda d : int(d.vmbus_id))
+vmbus_dev_list = sorted(vmbus_dev_list, key=lambda d: int(d.vmbus_id))
 
 format0 = '%2s: %s'
 format1 = '%2s: Class_ID = %s - %s\n%s'
@@ -95,9 +98,15 @@ for d in vmbus_dev_list:
 	if verbose == 0:
 		print(('VMBUS ID ' + format0) % (d.vmbus_id, d.dev_desc))
 	elif verbose == 1:
-		print (('VMBUS ID ' + format1) %	\
-			(d.vmbus_id, d.class_id, d.dev_desc, d.chn_vp_mapping))
+		print(
+			('VMBUS ID ' + format1) %
+			(d.vmbus_id, d.class_id, d.dev_desc, d.chn_vp_mapping)
+		)
 	else:
-		print (('VMBUS ID ' + format2) % \
-			(d.vmbus_id, d.class_id, d.dev_desc, \
-			d.device_id, d.sysfs_path, d.chn_vp_mapping))
+		print(
+			('VMBUS ID ' + format2) %
+			(
+				d.vmbus_id, d.class_id, d.dev_desc,
+				d.device_id, d.sysfs_path, d.chn_vp_mapping
+			)
+		)
Signature · 1 line
2.20.1
author maintainer
#45 lore raw

PATCH 44/44 tools: hv: fix KVP and VSS daemons exit code

In reply to Sasha Levin

From: Adrian Vladu <[email protected]>

[ Upstream commit b0995156071b0ff29a5902964a9dc8cfad6f81c0 ]

HyperV KVP and VSS daemons should exit with 0 when the '--help' or '-h' flags are used.

Cc "K. Y. Srinivasan" <[email protected]>
Cc Haiyang Zhang <[email protected]>
Cc Stephen Hemminger <[email protected]>
Cc Sasha Levin <[email protected]>
Cc Alessandro Pilotti <[email protected]>
---
 tools/hv/hv_kvp_daemon.c | 2 ++
 tools/hv/hv_vss_daemon.c | 2 ++
 2 files changed, 4 insertions(+)
tools/hv/hv_kvp_daemon.c +2 -0
index d7e06fe0270ee..0ce50c319cfd6 100644
@@ -1386,6 +1386,8 @@ int main(int argc, char *argv[])
 			daemonize = 0;
 			break;
 		case 'h':
+			print_usage(argv);
+			exit(0);
 		default:
 			print_usage(argv);
 			exit(EXIT_FAILURE);
tools/hv/hv_vss_daemon.c +2 -0
index efe1e34dd91b4..8f813f5233d48 100644
@@ -218,6 +218,8 @@ int main(int argc, char *argv[])
 			daemonize = 0;
 			break;
 		case 'h':
+			print_usage(argv);
+			exit(0);
 		default:
 			print_usage(argv);
 			exit(EXIT_FAILURE);
Signature · 1 line
2.20.1