Skip to contentSkip to conversation

5.18.18-rc1 review

PATCH In review Greg Kroah-Hartman 5 participants 1,101 messages Aug 15, 2022 – Aug 21, 2022 Page 9 of 9

Subsystems (12)
lore
Share & tools
lore
author maintainer
#1007 lore raw

PATCH 1005/1095 tracing/events: Add __vstring() and __assign_vstr() helper macros

In reply to Greg Kroah-Hartman

From: Steven Rostedt (Google) <[email protected]>

[ Upstream commit 0563231f93c6d1f582b168a47753b345c1e20d81 ]

There's several places that open code the following logic:

TP_STRUCT__entry(__dynamic_array(char, msg, MSG_MAX)), TP_fast_assign(vsnprintf(__get_str(msg), MSG_MAX, vaf->fmt, *vaf->va);)

To load a string created by variable array va_list.

The main issue with this approach is that "MSG_MAX" usage in the __dynamic_array() portion. That actually just reserves the MSG_MAX in the event, and even wastes space because there's dynamic meta data also saved in the event to denote the offset and size of the dynamic array. It would have been better to just use a static __array() field.

Instead, create __vstring() and __assign_vstr() that work like __string and __assign_str() but instead of taking a destination string to copy, take a format string and a va_list pointer and fill in the values.

It uses the helper:

#define __trace_event_vstr_len(fmt, va) \

 ({							\
	va_list __ap;					\
	int __ret;					\
							\
	va_copy(__ap, *(va));				\
	__ret = vsnprintf(NULL, 0, fmt, __ap) + 1;	\
	va_end(__ap);					\
							\
	min(__ret, TRACE_EVENT_STR_MAX);		\

})

To figure out the length to store the string. It may be slightly slower as it needs to run the vsnprintf() twice, but it now saves space on the ring buffer.

Cc Dennis Dalessandro <[email protected]>
Cc Ingo Molnar <[email protected]>
Cc Andrew Morton <[email protected]>
Cc Jason Gunthorpe <[email protected]>
Cc Leon Romanovsky <[email protected]>
Cc Kalle Valo <[email protected]>
Cc "David S. Miller" <[email protected]>
Cc Eric Dumazet <[email protected]>
Cc Jakub Kicinski <[email protected]>
Cc Paolo Abeni <[email protected]>
Cc Arend van Spriel <[email protected]>
Cc Franky Lin <[email protected]>
Cc Hante Meuleman <[email protected]>
Cc Gregory Greenman <[email protected]>
Cc Peter Chen <[email protected]>
Cc Greg Kroah-Hartman <[email protected]>
Cc Mathias Nyman <[email protected]>
Cc Chunfeng Yun <[email protected]>
Cc Bin Liu <[email protected]>
Cc Marek Lindner <[email protected]>
Cc Simon Wunderlich <[email protected]>
Cc Antonio Quartulli <[email protected]>
Cc Sven Eckelmann <[email protected]>
Cc Johannes Berg <[email protected]>
Cc Jim Cromie <[email protected]>
---
 include/linux/trace_events.h                 | 18 ++++++++++++++++++
 include/trace/stages/stage1_struct_define.h  |  3 +++
 include/trace/stages/stage2_data_offsets.h   |  3 +++
 include/trace/stages/stage4_event_fields.h   |  3 +++
 include/trace/stages/stage5_get_offsets.h    |  4 ++++
 include/trace/stages/stage6_event_callback.h |  7 +++++++
 6 files changed, 38 insertions(+)
include/linux/trace_events.h +18 -0
index e6e95a9f07a5..b18759a673c6 100644
@@ -916,6 +916,24 @@ perf_trace_buf_submit(void *raw_data, int size, int rctx, u16 type,
 
 #endif
 
+#define TRACE_EVENT_STR_MAX	512
+
+/*
+ * gcc warns that you can not use a va_list in an inlined
+ * function. But lets me make it into a macro :-/
+ */
+#define __trace_event_vstr_len(fmt, va)			\
+({							\
+	va_list __ap;					\
+	int __ret;					\
+							\
+	va_copy(__ap, *(va));				\
+	__ret = vsnprintf(NULL, 0, fmt, __ap) + 1;	\
+	va_end(__ap);					\
+							\
+	min(__ret, TRACE_EVENT_STR_MAX);		\
+})
+
 #endif /* _LINUX_TRACE_EVENT_H */
 
 /*
include/trace/stages/stage1_struct_define.h +3 -0
index a16783419687..1b7bab60434c 100644
@@ -26,6 +26,9 @@
 #undef __string_len
 #define __string_len(item, src, len) __dynamic_array(char, item, -1)
 
+#undef __vstring
+#define __vstring(item, fmt, ap) __dynamic_array(char, item, -1)
+
 #undef __bitmask
 #define __bitmask(item, nr_bits) __dynamic_array(char, item, -1)
 
include/trace/stages/stage2_data_offsets.h +3 -0
index 42fd1e8813ec..1b7a8f764fdd 100644
@@ -32,6 +32,9 @@
 #undef __string_len
 #define __string_len(item, src, len) __dynamic_array(char, item, -1)
 
+#undef __vstring
+#define __vstring(item, fmt, ap) __dynamic_array(char, item, -1)
+
 #undef __bitmask
 #define __bitmask(item, nr_bits) __dynamic_array(unsigned long, item, -1)
 
include/trace/stages/stage4_event_fields.h +3 -0
index e80cdc397a43..c3790ec7a453 100644
@@ -38,6 +38,9 @@
 #undef __string_len
 #define __string_len(item, src, len) __dynamic_array(char, item, -1)
 
+#undef __vstring
+#define __vstring(item, fmt, ap) __dynamic_array(char, item, -1)
+
 #undef __bitmask
 #define __bitmask(item, nr_bits) __dynamic_array(unsigned long, item, -1)
 
include/trace/stages/stage5_get_offsets.h +4 -0
index 7ee5931300e6..fba4c24ed9e6 100644
@@ -39,6 +39,10 @@
 #undef __string_len
 #define __string_len(item, src, len) __dynamic_array(char, item, (len) + 1)
 
+#undef __vstring
+#define __vstring(item, fmt, ap) __dynamic_array(char, item,		\
+		      __trace_event_vstr_len(fmt, ap))
+
 #undef __rel_dynamic_array
 #define __rel_dynamic_array(type, item, len)				\
 	__item_length = (len) * sizeof(type);				\
include/trace/stages/stage6_event_callback.h +7 -0
index e1724f73594b..0f51f6b3ab70 100644
@@ -24,6 +24,9 @@
 #undef __string_len
 #define __string_len(item, src, len) __dynamic_array(char, item, -1)
 
+#undef __vstring
+#define __vstring(item, fmt, ap) __dynamic_array(char, item, -1)
+
 #undef __assign_str
 #define __assign_str(dst, src)						\
 	strcpy(__get_str(dst), (src) ? (const char *)(src) : "(null)");
@@ -35,6 +38,10 @@
 		__get_str(dst)[len] = '\0';				\
 	} while(0)
 
+#undef __assign_vstr
+#define __assign_vstr(dst, fmt, va)					\
+	vsnprintf(__get_str(dst), TRACE_EVENT_STR_MAX, fmt, *(va))
+
 #undef __bitmask
 #define __bitmask(item, nr_bits) __dynamic_array(unsigned long, item, -1)
 
Signature · 1 line
2.35.1
author maintainer
#1008 lore raw

PATCH 1006/1095 dm thin: fix use-after-free crash in dm_sm_register_threshold_callback

In reply to Greg Kroah-Hartman

From: Luo Meng <[email protected]>

[ Upstream commit 3534e5a5ed2997ca1b00f44a0378a075bd05e8a3 ]

Fault inject on pool metadata device reports:

  BUG: KASAN: use-after-free in dm_pool_register_metadata_threshold+0x40/0x80
  Read of size 8 at addr ffff8881b9d50068 by task dmsetup/950

  CPU: 7 PID: 950 Comm: dmsetup Tainted: G        W         5.19.0-rc6 #1
  Hardware name: QEMU Standard PC (i440FX + PIIX, 1996), BIOS 1.14.0-1.fc33 04/01/2014
  Call Trace:
   <TASK>
   dump_stack_lvl+0x34/0x44
   print_address_description.constprop.0.cold+0xeb/0x3f4
   kasan_report.cold+0xe6/0x147
   dm_pool_register_metadata_threshold+0x40/0x80
   pool_ctr+0xa0a/0x1150
   dm_table_add_target+0x2c8/0x640
   table_load+0x1fd/0x430
   ctl_ioctl+0x2c4/0x5a0
   dm_ctl_ioctl+0xa/0x10
   __x64_sys_ioctl+0xb3/0xd0
   do_syscall_64+0x35/0x80
   entry_SYSCALL_64_after_hwframe+0x46/0xb0

This can be easily reproduced using:
echo offline > /sys/block/sda/device/state
dd if=/dev/zero of=/dev/mapper/thin bs=4k count=10
dmsetup load pool --table "0 20971520 thin-pool /dev/sda /dev/sdb 128 0 0"

If a metadata commit fails, the transaction will be aborted and the metadata space maps will be destroyed. If a DM table reload then happens for this failed thin-pool, a use-after-free will occur in dm_sm_register_threshold_callback (called from dm_pool_register_metadata_threshold).

Fix this by in dm_pool_register_metadata_threshold() by returning the -EINVAL error if the thin-pool is in fail mode. Also fail pool_ctr() with a new error message: "Error registering metadata threshold".

Fixes ac8c3f3df65e4 ("dm thin: generate event when metadata threshold passed")
---
 drivers/md/dm-thin-metadata.c | 7 +++++--
 drivers/md/dm-thin.c          | 4 +++-
 2 files changed, 8 insertions(+), 3 deletions(-)
drivers/md/dm-thin-metadata.c +5 -2
index 2db7030aba00..a27395c8621f 100644
@@ -2045,10 +2045,13 @@ int dm_pool_register_metadata_threshold(struct dm_pool_metadata *pmd,
 					dm_sm_threshold_fn fn,
 					void *context)
 {
-	int r;
+	int r = -EINVAL;
 
 	pmd_write_lock_in_core(pmd);
-	r = dm_sm_register_threshold_callback(pmd->metadata_sm, threshold, fn, context);
+	if (!pmd->fail_io) {
+		r = dm_sm_register_threshold_callback(pmd->metadata_sm,
+						      threshold, fn, context);
+	}
 	pmd_write_unlock(pmd);
 
 	return r;
drivers/md/dm-thin.c +3 -1
index 4d25d0e27031..53ac6ae870ac 100644
@@ -3382,8 +3382,10 @@ static int pool_ctr(struct dm_target *ti, unsigned argc, char **argv)
 						calc_metadata_threshold(pt),
 						metadata_low_callback,
 						pool);
-	if (r)
+	if (r) {
+		ti->error = "Error registering metadata threshold";
 		goto out_flags_changed;
+	}
 
 	dm_pool_register_pre_commit_callback(pool->pmd,
 					     metadata_pre_commit_callback, pool);
Signature · 1 line
2.35.1
author maintainer
#1009 lore raw

PATCH 1007/1095 net/9p: Initialize the iounit field during fid creation

In reply to Greg Kroah-Hartman

From: Tyler Hicks <[email protected]>

[ Upstream commit aa7aeee169480e98cf41d83c01290a37e569be6d ]

Ensure that the fid's iounit field is set to zero when a new fid is created. Certain 9P operations, such as OPEN and CREATE, allow the server to reply with an iounit size which the client code assigns to the p9_fid struct shortly after the fid is created by p9_fid_create(). On the other hand, an XATTRWALK operation doesn't allow for the server to specify an iounit value. The iounit field of the newly allocated p9_fid struct remained uninitialized in that case. Depending on allocation patterns, the iounit value could have been something reasonable that was carried over from previously freed fids or, in the worst case, could have been arbitrary values from non-fid related usages of the memory location.

The bug was detected in the Windows Subsystem for Linux 2 (WSL2) kernel after the uninitialized iounit field resulted in the typical sequence of two getxattr(2) syscalls, one to get the size of an xattr and another after allocating a sufficiently sized buffer to fit the xattr value, to hit an unexpected ERANGE error in the second call to getxattr(2). An uninitialized iounit field would sometimes force rsize to be smaller than the xattr value size in p9_client_read_once() and the 9P server in WSL refused to chunk up the READ on the attr_fid and, instead, returned ERANGE to the client. The virtfs server in QEMU seems happy to chunk up the READ and this problem goes undetected there.

---
 net/9p/client.c | 5 +----
 1 file changed, 1 insertion(+), 4 deletions(-)
net/9p/client.c +1 -4
index a36a40137caa..87cde948f628 100644
@@ -886,16 +886,13 @@ static struct p9_fid *p9_fid_create(struct p9_client *clnt)
 	struct p9_fid *fid;
 
 	p9_debug(P9_DEBUG_FID, "clnt %p\n", clnt);
-	fid = kmalloc(sizeof(*fid), GFP_KERNEL);
+	fid = kzalloc(sizeof(*fid), GFP_KERNEL);
 	if (!fid)
 		return NULL;
 
-	memset(&fid->qid, 0, sizeof(fid->qid));
 	fid->mode = -1;
 	fid->uid = current_fsuid();
 	fid->clnt = clnt;
-	fid->rdir = NULL;
-	fid->fid = 0;
 	refcount_set(&fid->count, 1);
 
 	idr_preload(GFP_KERNEL);
Signature · 1 line
2.35.1
author maintainer
#1010 lore raw

PATCH 1008/1095 timekeeping: contribute wall clock to rng on time change

In reply to Greg Kroah-Hartman

From: Jason A. Donenfeld <[email protected]>

[ Upstream commit b8ac29b40183a6038919768b5d189c9bd91ce9b4 ]

The rng's random_init() function contributes the real time to the rng at boot time, so that events can at least start in relation to something particular in the real world. But this clock might not yet be set that point in boot, so nothing is contributed. In addition, the relation between minor clock changes from, say, NTP, and the cycle counter is potentially useful entropic data.

This commit addresses this by mixing in a time stamp on calls to settimeofday and adjtimex. No entropy is credited in doing so, so it doesn't make initialization faster, but it is still useful input to have.

---
 kernel/time/timekeeping.c | 7 ++++++-
 1 file changed, 6 insertions(+), 1 deletion(-)
kernel/time/timekeeping.c +6 -1
index 871c912860ed..d6a0ff68df41 100644
@@ -23,6 +23,7 @@
 #include <linux/pvclock_gtod.h>
 #include <linux/compiler.h>
 #include <linux/audit.h>
+#include <linux/random.h>
 
 #include "tick-internal.h"
 #include "ntp_internal.h"
@@ -1326,8 +1327,10 @@ int do_settimeofday64(const struct timespec64 *ts)
 	/* Signal hrtimers about time change */
 	clock_was_set(CLOCK_SET_WALL);
 
-	if (!ret)
+	if (!ret) {
 		audit_tk_injoffset(ts_delta);
+		add_device_randomness(ts, sizeof(*ts));
+	}
 
 	return ret;
 }
@@ -2413,6 +2416,7 @@ int do_adjtimex(struct __kernel_timex *txc)
 	ret = timekeeping_validate_timex(txc);
 	if (ret)
 		return ret;
+	add_device_randomness(txc, sizeof(*txc));
 
 	if (txc->modes & ADJ_SETOFFSET) {
 		struct timespec64 delta;
@@ -2430,6 +2434,7 @@ int do_adjtimex(struct __kernel_timex *txc)
 	audit_ntp_init(&ad);
 
 	ktime_get_real_ts64(&ts);
+	add_device_randomness(&ts, sizeof(ts));
 
 	raw_spin_lock_irqsave(&timekeeper_lock, flags);
 	write_seqcount_begin(&tk_core.seq);
Signature · 1 line
2.35.1
author maintainer
#1011 lore raw

PATCH 1009/1095 locking/csd_lock: Change csdlock_debug from early_param to __setup

In reply to Greg Kroah-Hartman

From: Chen Zhongjin <[email protected]>

[ Upstream commit 9c9b26b0df270d4f9246e483a44686fca951a29c ]

The csdlock_debug kernel-boot parameter is parsed by the early_param() function csdlock_debug(). If set, csdlock_debug() invokes static_branch_enable() to enable csd_lock_wait feature, which triggers a panic on arm64 for kernels built with CONFIG_SPARSEMEM=y and CONFIG_SPARSEMEM_VMEMMAP=n.

With CONFIG_SPARSEMEM_VMEMMAP=n, __nr_to_section is called in static_key_enable() and returns NULL, resulting in a NULL dereference because mem_section is initialized only later in sparse_init().

This is also a problem for powerpc because early_param() functions are invoked earlier than jump_label_init(), also resulting in static_key_enable() failures. These failures cause the warning "static key 'xxx' used before call to jump_label_init()".

Thus, early_param is too early for csd_lock_wait to run static_branch_enable(), so changes it to __setup to fix these.

Fixes 8d0968cc6b8f ("locking/csd_lock: Add boot parameter for controlling CSD lock debugging")
---
 kernel/smp.c | 4 ++--
 1 file changed, 2 insertions(+), 2 deletions(-)
kernel/smp.c +2 -2
index 65a630f62363..381eb15cd28f 100644
@@ -174,9 +174,9 @@ static int __init csdlock_debug(char *str)
 	if (val)
 		static_branch_enable(&csdlock_debug_enabled);
 
-	return 0;
+	return 1;
 }
-early_param("csdlock_debug", csdlock_debug);
+__setup("csdlock_debug=", csdlock_debug);
 
 static DEFINE_PER_CPU(call_single_data_t *, cur_csd);
 static DEFINE_PER_CPU(smp_call_func_t, cur_csd_func);
Signature · 1 line
2.35.1
author maintainer
#1012 lore raw

PATCH 1010/1095 block: serialize all debugfs operations using q->debugfs_mutex

In reply to Greg Kroah-Hartman

From: Christoph Hellwig <[email protected]>

[ Upstream commit 5cf9c91ba927119fc6606b938b1895bb2459d3bc ]

Various places like I/O schedulers or the QOS infrastructure try to register debugfs files on demans, which can race with creating and removing the main queue debugfs directory. Use the existing debugfs_mutex to serialize all debugfs operations that rely on q->debugfs_dir or the directories hanging off it.

To make the teardown code a little simpler declare all debugfs dentry pointers and not just the main one uncoditionally in blkdev.h.

Move debugfs_mutex next to the dentries that it protects and document what it is used for.

---
 block/blk-mq-debugfs.c  | 25 ++++++++++++++++++++-----
 block/blk-mq-debugfs.h  |  5 -----
 block/blk-mq-sched.c    | 11 +++++++++++
 block/blk-rq-qos.c      |  2 ++
 block/blk-rq-qos.h      |  7 ++++++-
 block/blk-sysfs.c       | 20 +++++++++-----------
 include/linux/blkdev.h  |  8 ++++----
 kernel/trace/blktrace.c |  3 ---
 8 files changed, 52 insertions(+), 29 deletions(-)
block/blk-mq-debugfs.c +20 -5
index 34bee263936c..d491b6eb0ab9 100644
@@ -713,11 +713,6 @@ void blk_mq_debugfs_register(struct request_queue *q)
 	}
 }
 
-void blk_mq_debugfs_unregister(struct request_queue *q)
-{
-	q->sched_debugfs_dir = NULL;
-}
-
 static void blk_mq_debugfs_register_ctx(struct blk_mq_hw_ctx *hctx,
 					struct blk_mq_ctx *ctx)
 {
@@ -751,6 +746,8 @@ void blk_mq_debugfs_register_hctx(struct request_queue *q,
 
 void blk_mq_debugfs_unregister_hctx(struct blk_mq_hw_ctx *hctx)
 {
+	if (!hctx->queue->debugfs_dir)
+		return;
 	debugfs_remove_recursive(hctx->debugfs_dir);
 	hctx->sched_debugfs_dir = NULL;
 	hctx->debugfs_dir = NULL;
@@ -778,6 +775,8 @@ void blk_mq_debugfs_register_sched(struct request_queue *q)
 {
 	struct elevator_type *e = q->elevator->type;
 
+	lockdep_assert_held(&q->debugfs_mutex);
+
 	/*
 	 * If the parent directory has not been created yet, return, we will be
 	 * called again later on and the directory/files will be created then.
@@ -795,6 +794,8 @@ void blk_mq_debugfs_register_sched(struct request_queue *q)
 
 void blk_mq_debugfs_unregister_sched(struct request_queue *q)
 {
+	lockdep_assert_held(&q->debugfs_mutex);
+
 	debugfs_remove_recursive(q->sched_debugfs_dir);
 	q->sched_debugfs_dir = NULL;
 }
@@ -816,6 +817,10 @@ static const char *rq_qos_id_to_name(enum rq_qos_id id)
 
 void blk_mq_debugfs_unregister_rqos(struct rq_qos *rqos)
 {
+	lockdep_assert_held(&rqos->q->debugfs_mutex);
+
+	if (!rqos->q->debugfs_dir)
+		return;
 	debugfs_remove_recursive(rqos->debugfs_dir);
 	rqos->debugfs_dir = NULL;
 }
@@ -825,6 +830,8 @@ void blk_mq_debugfs_register_rqos(struct rq_qos *rqos)
 	struct request_queue *q = rqos->q;
 	const char *dir_name = rq_qos_id_to_name(rqos->id);
 
+	lockdep_assert_held(&q->debugfs_mutex);
+
 	if (rqos->debugfs_dir || !rqos->ops->debugfs_attrs)
 		return;
 
@@ -840,6 +847,8 @@ void blk_mq_debugfs_register_rqos(struct rq_qos *rqos)
 
 void blk_mq_debugfs_unregister_queue_rqos(struct request_queue *q)
 {
+	lockdep_assert_held(&q->debugfs_mutex);
+
 	debugfs_remove_recursive(q->rqos_debugfs_dir);
 	q->rqos_debugfs_dir = NULL;
 }
@@ -849,6 +858,8 @@ void blk_mq_debugfs_register_sched_hctx(struct request_queue *q,
 {
 	struct elevator_type *e = q->elevator->type;
 
+	lockdep_assert_held(&q->debugfs_mutex);
+
 	/*
 	 * If the parent debugfs directory has not been created yet, return;
 	 * We will be called again later on with appropriate parent debugfs
@@ -868,6 +879,10 @@ void blk_mq_debugfs_register_sched_hctx(struct request_queue *q,
 
 void blk_mq_debugfs_unregister_sched_hctx(struct blk_mq_hw_ctx *hctx)
 {
+	lockdep_assert_held(&hctx->queue->debugfs_mutex);
+
+	if (!hctx->queue->debugfs_dir)
+		return;
 	debugfs_remove_recursive(hctx->sched_debugfs_dir);
 	hctx->sched_debugfs_dir = NULL;
 }
block/blk-mq-debugfs.h +0 -5
index 69918f4170d6..771d45832878 100644
@@ -21,7 +21,6 @@ int __blk_mq_debugfs_rq_show(struct seq_file *m, struct request *rq);
 int blk_mq_debugfs_rq_show(struct seq_file *m, void *v);
 
 void blk_mq_debugfs_register(struct request_queue *q);
-void blk_mq_debugfs_unregister(struct request_queue *q);
 void blk_mq_debugfs_register_hctx(struct request_queue *q,
 				  struct blk_mq_hw_ctx *hctx);
 void blk_mq_debugfs_unregister_hctx(struct blk_mq_hw_ctx *hctx);
@@ -42,10 +41,6 @@ static inline void blk_mq_debugfs_register(struct request_queue *q)
 {
 }
 
-static inline void blk_mq_debugfs_unregister(struct request_queue *q)
-{
-}
-
 static inline void blk_mq_debugfs_register_hctx(struct request_queue *q,
 						struct blk_mq_hw_ctx *hctx)
 {
block/blk-mq-sched.c +11 -0
index 9e56a69422b6..e84bec39fd3a 100644
@@ -593,7 +593,9 @@ int blk_mq_init_sched(struct request_queue *q, struct elevator_type *e)
 	if (ret)
 		goto err_free_map_and_rqs;
 
+	mutex_lock(&q->debugfs_mutex);
 	blk_mq_debugfs_register_sched(q);
+	mutex_unlock(&q->debugfs_mutex);
 
 	queue_for_each_hw_ctx(q, hctx, i) {
 		if (e->ops.init_hctx) {
@@ -606,7 +608,9 @@ int blk_mq_init_sched(struct request_queue *q, struct elevator_type *e)
 				return ret;
 			}
 		}
+		mutex_lock(&q->debugfs_mutex);
 		blk_mq_debugfs_register_sched_hctx(q, hctx);
+		mutex_unlock(&q->debugfs_mutex);
 	}
 
 	return 0;
@@ -647,14 +651,21 @@ void blk_mq_exit_sched(struct request_queue *q, struct elevator_queue *e)
 	unsigned int flags = 0;
 
 	queue_for_each_hw_ctx(q, hctx, i) {
+		mutex_lock(&q->debugfs_mutex);
 		blk_mq_debugfs_unregister_sched_hctx(hctx);
+		mutex_unlock(&q->debugfs_mutex);
+
 		if (e->type->ops.exit_hctx && hctx->sched_data) {
 			e->type->ops.exit_hctx(hctx, i);
 			hctx->sched_data = NULL;
 		}
 		flags = hctx->flags;
 	}
+
+	mutex_lock(&q->debugfs_mutex);
 	blk_mq_debugfs_unregister_sched(q);
+	mutex_unlock(&q->debugfs_mutex);
+
 	if (e->type->ops.exit_sched)
 		e->type->ops.exit_sched(e);
 	blk_mq_sched_tags_teardown(q, flags);
block/blk-rq-qos.c +2 -0
index e83af7bc7591..249a6f05dd3b 100644
@@ -294,7 +294,9 @@ void rq_qos_wait(struct rq_wait *rqw, void *private_data,
 
 void rq_qos_exit(struct request_queue *q)
 {
+	mutex_lock(&q->debugfs_mutex);
 	blk_mq_debugfs_unregister_queue_rqos(q);
+	mutex_unlock(&q->debugfs_mutex);
 
 	while (q->rq_qos) {
 		struct rq_qos *rqos = q->rq_qos;
block/blk-rq-qos.h +6 -1
index 68267007da1c..0e46052b018a 100644
@@ -104,8 +104,11 @@ static inline void rq_qos_add(struct request_queue *q, struct rq_qos *rqos)
 
 	blk_mq_unfreeze_queue(q);
 
-	if (rqos->ops->debugfs_attrs)
+	if (rqos->ops->debugfs_attrs) {
+		mutex_lock(&q->debugfs_mutex);
 		blk_mq_debugfs_register_rqos(rqos);
+		mutex_unlock(&q->debugfs_mutex);
+	}
 }
 
 static inline void rq_qos_del(struct request_queue *q, struct rq_qos *rqos)
@@ -129,7 +132,9 @@ static inline void rq_qos_del(struct request_queue *q, struct rq_qos *rqos)
 
 	blk_mq_unfreeze_queue(q);
 
+	mutex_lock(&q->debugfs_mutex);
 	blk_mq_debugfs_unregister_rqos(rqos);
+	mutex_unlock(&q->debugfs_mutex);
 }
 
 typedef bool (acquire_inflight_cb_t)(struct rq_wait *rqw, void *private_data);
block/blk-sysfs.c +9 -11
index 88bd41d4cb59..6e4801b217a7 100644
@@ -779,14 +779,13 @@ static void blk_release_queue(struct kobject *kobj)
 	if (queue_is_mq(q))
 		blk_mq_release(q);
 
-	blk_trace_shutdown(q);
 	mutex_lock(&q->debugfs_mutex);
+	blk_trace_shutdown(q);
 	debugfs_remove_recursive(q->debugfs_dir);
+	q->debugfs_dir = NULL;
+	q->sched_debugfs_dir = NULL;
 	mutex_unlock(&q->debugfs_mutex);
 
-	if (queue_is_mq(q))
-		blk_mq_debugfs_unregister(q);
-
 	bioset_exit(&q->bio_split);
 
 	if (blk_queue_has_srcu(q))
@@ -836,17 +835,16 @@ int blk_register_queue(struct gendisk *disk)
 		goto unlock;
 	}
 
+	if (queue_is_mq(q))
+		__blk_mq_register_dev(dev, q);
+	mutex_lock(&q->sysfs_lock);
+
 	mutex_lock(&q->debugfs_mutex);
 	q->debugfs_dir = debugfs_create_dir(kobject_name(q->kobj.parent),
 					    blk_debugfs_root);
-	mutex_unlock(&q->debugfs_mutex);
-
-	if (queue_is_mq(q)) {
-		__blk_mq_register_dev(dev, q);
+	if (queue_is_mq(q))
 		blk_mq_debugfs_register(q);
-	}
-
-	mutex_lock(&q->sysfs_lock);
+	mutex_unlock(&q->debugfs_mutex);
 
 	ret = disk_register_independent_access_ranges(disk, NULL);
 	if (ret)
include/linux/blkdev.h +4 -4
index 108e3d114bfc..cc6b24a5098f 100644
@@ -466,7 +466,6 @@ struct request_queue {
 #endif /* CONFIG_BLK_DEV_ZONED */
 
 	int			node;
-	struct mutex		debugfs_mutex;
 #ifdef CONFIG_BLK_DEV_IO_TRACE
 	struct blk_trace __rcu	*blk_trace;
 #endif
@@ -510,11 +509,12 @@ struct request_queue {
 	struct bio_set		bio_split;
 
 	struct dentry		*debugfs_dir;
-
-#ifdef CONFIG_BLK_DEBUG_FS
 	struct dentry		*sched_debugfs_dir;
 	struct dentry		*rqos_debugfs_dir;
-#endif
+	/*
+	 * Serializes all debugfs metadata operations using the above dentries.
+	 */
+	struct mutex		debugfs_mutex;
 
 	bool			mq_sysfs_init_done;
 
kernel/trace/blktrace.c +0 -3
index f22219495541..f0500b5cfefe 100644
@@ -770,14 +770,11 @@ int blk_trace_ioctl(struct block_device *bdev, unsigned cmd, char __user *arg)
  **/
 void blk_trace_shutdown(struct request_queue *q)
 {
-	mutex_lock(&q->debugfs_mutex);
 	if (rcu_dereference_protected(q->blk_trace,
 				      lockdep_is_held(&q->debugfs_mutex))) {
 		__blk_trace_startstop(q, 0);
 		__blk_trace_remove(q);
 	}
-
-	mutex_unlock(&q->debugfs_mutex);
 }
 
 #ifdef CONFIG_BLK_CGROUP
Signature · 1 line
2.35.1
author maintainer
#1013 lore raw

PATCH 1011/1095 block: dont allow the same type rq_qos add more than once

In reply to Greg Kroah-Hartman

From: Jinke Han <[email protected]>

[ Upstream commit 14a6e2eb7df5c7897c15b109cba29ab0c4a791b6 ]

In our test of iocost, we encountered some list add/del corruptions of inner_walk list in ioc_timer_fn.

The reason can be described as follows:

cpu 0 cpu 1
ioc_qos_write ioc_qos_write

ioc = q_to_ioc(queue);

if (!ioc) {
        ioc = kzalloc();
					ioc = q_to_ioc(queue);
					if (!ioc) {
						ioc = kzalloc();
						...
						rq_qos_add(q, rqos);
					}
        ...
        rq_qos_add(q, rqos);
        ...
}

When the io.cost.qos file is written by two cpus concurrently, rq_qos may be added to one disk twice. In that case, there will be two iocs enabled and running on one disk. They own different iocgs on their active list. In the ioc_timer_fn function, because of the iocgs from two iocs have the same root iocg, the root iocg's walk_list may be overwritten by each other and this leads to list add/del corruptions in building or destroying the inner_walk list.

And so far, the blk-rq-qos framework works in case that one instance for one type rq_qos per queue by default. This patch make this explicit and also fix the crash above.

---
 block/blk-iocost.c    | 20 +++++++++++++-------
 block/blk-iolatency.c | 18 +++++++++++-------
 block/blk-rq-qos.h    | 11 ++++++++++-
 block/blk-wbt.c       | 12 +++++++++++-
 4 files changed, 45 insertions(+), 16 deletions(-)
block/blk-iocost.c +13 -7
index 16705fbd0699..a19f2db4eeb2 100644
@@ -2893,15 +2893,21 @@ static int blk_iocost_init(struct request_queue *q)
 	 * called before policy activation completion, can't assume that the
 	 * target bio has an iocg associated and need to test for NULL iocg.
 	 */
-	rq_qos_add(q, rqos);
+	ret = rq_qos_add(q, rqos);
+	if (ret)
+		goto err_free_ioc;
+
 	ret = blkcg_activate_policy(q, &blkcg_policy_iocost);
-	if (ret) {
-		rq_qos_del(q, rqos);
-		free_percpu(ioc->pcpu_stat);
-		kfree(ioc);
-		return ret;
-	}
+	if (ret)
+		goto err_del_qos;
 	return 0;
+
+err_del_qos:
+	rq_qos_del(q, rqos);
+err_free_ioc:
+	free_percpu(ioc->pcpu_stat);
+	kfree(ioc);
+	return ret;
 }
 
 static struct blkcg_policy_data *ioc_cpd_alloc(gfp_t gfp)
block/blk-iolatency.c +11 -7
index 9568bf8dfe82..7845dca5fcfd 100644
@@ -773,19 +773,23 @@ int blk_iolatency_init(struct request_queue *q)
 	rqos->ops = &blkcg_iolatency_ops;
 	rqos->q = q;
 
-	rq_qos_add(q, rqos);
-
+	ret = rq_qos_add(q, rqos);
+	if (ret)
+		goto err_free;
 	ret = blkcg_activate_policy(q, &blkcg_policy_iolatency);
-	if (ret) {
-		rq_qos_del(q, rqos);
-		kfree(blkiolat);
-		return ret;
-	}
+	if (ret)
+		goto err_qos_del;
 
 	timer_setup(&blkiolat->timer, blkiolatency_timer_fn, 0);
 	INIT_WORK(&blkiolat->enable_work, blkiolatency_enable_work_fn);
 
 	return 0;
+
+err_qos_del:
+	rq_qos_del(q, rqos);
+err_free:
+	kfree(blkiolat);
+	return ret;
 }
 
 static void iolatency_set_min_lat_nsec(struct blkcg_gq *blkg, u64 val)
block/blk-rq-qos.h +10 -1
index 0e46052b018a..08b856570ad1 100644
@@ -86,7 +86,7 @@ static inline void rq_wait_init(struct rq_wait *rq_wait)
 	init_waitqueue_head(&rq_wait->wait);
 }
 
-static inline void rq_qos_add(struct request_queue *q, struct rq_qos *rqos)
+static inline int rq_qos_add(struct request_queue *q, struct rq_qos *rqos)
 {
 	/*
 	 * No IO can be in-flight when adding rqos, so freeze queue, which
@@ -98,6 +98,8 @@ static inline void rq_qos_add(struct request_queue *q, struct rq_qos *rqos)
 	blk_mq_freeze_queue(q);
 
 	spin_lock_irq(&q->queue_lock);
+	if (rq_qos_id(q, rqos->id))
+		goto ebusy;
 	rqos->next = q->rq_qos;
 	q->rq_qos = rqos;
 	spin_unlock_irq(&q->queue_lock);
@@ -109,6 +111,13 @@ static inline void rq_qos_add(struct request_queue *q, struct rq_qos *rqos)
 		blk_mq_debugfs_register_rqos(rqos);
 		mutex_unlock(&q->debugfs_mutex);
 	}
+
+	return 0;
+ebusy:
+	spin_unlock_irq(&q->queue_lock);
+	blk_mq_unfreeze_queue(q);
+	return -EBUSY;
+
 }
 
 static inline void rq_qos_del(struct request_queue *q, struct rq_qos *rqos)
block/blk-wbt.c +11 -1
index 0c119be0e813..ae6ea0b54579 100644
@@ -820,6 +820,7 @@ int wbt_init(struct request_queue *q)
 {
 	struct rq_wb *rwb;
 	int i;
+	int ret;
 
 	rwb = kzalloc(sizeof(*rwb), GFP_KERNEL);
 	if (!rwb)
@@ -846,7 +847,10 @@ int wbt_init(struct request_queue *q)
 	/*
 	 * Assign rwb and add the stats callback.
 	 */
-	rq_qos_add(q, &rwb->rqos);
+	ret = rq_qos_add(q, &rwb->rqos);
+	if (ret)
+		goto err_free;
+
 	blk_stat_add_callback(q, rwb->cb);
 
 	rwb->min_lat_nsec = wbt_default_latency_nsec(q);
@@ -855,4 +859,10 @@ int wbt_init(struct request_queue *q)
 	wbt_set_write_cache(q, test_bit(QUEUE_FLAG_WC, &q->queue_flags));
 
 	return 0;
+
+err_free:
+	blk_stat_free_callback(rwb->cb);
+	kfree(rwb);
+	return ret;
+
 }
Signature · 1 line
2.35.1
author maintainer
#1014 lore raw

PATCH 1012/1095 btrfs: tree-log: make the return value for log syncing consistent

In reply to Greg Kroah-Hartman

From: Josef Bacik <[email protected]>

[ Upstream commit f31f09f6be1c6c1a673e0566e258281a7bbaaa51 ]

Currently we will return 1 or -EAGAIN if we decide we need to commit the transaction rather than sync the log. In practice this doesn't really matter, we interpret any !0 and !BTRFS_NO_LOG_SYNC as needing to commit the transaction. However this makes it hard to figure out what the correct thing to do is.

Fix this up by defining BTRFS_LOG_FORCE_COMMIT and using this in all the places where we want to force the transaction to be committed.

---
 fs/btrfs/file.c     |  2 +-
 fs/btrfs/tree-log.c | 18 +++++++++---------
 fs/btrfs/tree-log.h |  3 +++
 3 files changed, 13 insertions(+), 10 deletions(-)
fs/btrfs/file.c +1 -1
index 153920acd226..2d24f2dcc0ea 100644
@@ -2344,7 +2344,7 @@ int btrfs_sync_file(struct file *file, loff_t start, loff_t end, int datasync)
 	btrfs_release_log_ctx_extents(&ctx);
 	if (ret < 0) {
 		/* Fallthrough and commit/free transaction. */
-		ret = 1;
+		ret = BTRFS_LOG_FORCE_COMMIT;
 	}
 
 	/* we've logged all the items and now have a consistent
fs/btrfs/tree-log.c +9 -9
index e65633686378..08917069a125 100644
@@ -171,7 +171,7 @@ static int start_log_trans(struct btrfs_trans_handle *trans,
 		int index = (root->log_transid + 1) % 2;
 
 		if (btrfs_need_log_full_commit(trans)) {
-			ret = -EAGAIN;
+			ret = BTRFS_LOG_FORCE_COMMIT;
 			goto out;
 		}
 
@@ -194,7 +194,7 @@ static int start_log_trans(struct btrfs_trans_handle *trans,
 		 * writing.
 		 */
 		if (zoned && !created) {
-			ret = -EAGAIN;
+			ret = BTRFS_LOG_FORCE_COMMIT;
 			goto out;
 		}
 
@@ -3122,7 +3122,7 @@ int btrfs_sync_log(struct btrfs_trans_handle *trans,
 
 	/* bail out if we need to do a full commit */
 	if (btrfs_need_log_full_commit(trans)) {
-		ret = -EAGAIN;
+		ret = BTRFS_LOG_FORCE_COMMIT;
 		mutex_unlock(&root->log_mutex);
 		goto out;
 	}
@@ -3223,7 +3223,7 @@ int btrfs_sync_log(struct btrfs_trans_handle *trans,
 		}
 		btrfs_wait_tree_log_extents(log, mark);
 		mutex_unlock(&log_root_tree->log_mutex);
-		ret = -EAGAIN;
+		ret = BTRFS_LOG_FORCE_COMMIT;
 		goto out;
 	}
 
@@ -3262,7 +3262,7 @@ int btrfs_sync_log(struct btrfs_trans_handle *trans,
 		blk_finish_plug(&plug);
 		btrfs_wait_tree_log_extents(log, mark);
 		mutex_unlock(&log_root_tree->log_mutex);
-		ret = -EAGAIN;
+		ret = BTRFS_LOG_FORCE_COMMIT;
 		goto out_wake_log_root;
 	}
 
@@ -5849,7 +5849,7 @@ static int btrfs_log_inode(struct btrfs_trans_handle *trans,
 	    inode_only == LOG_INODE_ALL &&
 	    inode->last_unlink_trans >= trans->transid) {
 		btrfs_set_log_full_commit(trans);
-		ret = 1;
+		ret = BTRFS_LOG_FORCE_COMMIT;
 		goto out_unlock;
 	}
 
@@ -6563,12 +6563,12 @@ static int btrfs_log_inode_parent(struct btrfs_trans_handle *trans,
 	bool log_dentries = false;
 
 	if (btrfs_test_opt(fs_info, NOTREELOG)) {
-		ret = 1;
+		ret = BTRFS_LOG_FORCE_COMMIT;
 		goto end_no_trans;
 	}
 
 	if (btrfs_root_refs(&root->root_item) == 0) {
-		ret = 1;
+		ret = BTRFS_LOG_FORCE_COMMIT;
 		goto end_no_trans;
 	}
 
@@ -6666,7 +6666,7 @@ static int btrfs_log_inode_parent(struct btrfs_trans_handle *trans,
 end_trans:
 	if (ret < 0) {
 		btrfs_set_log_full_commit(trans);
-		ret = 1;
+		ret = BTRFS_LOG_FORCE_COMMIT;
 	}
 
 	if (ret)
fs/btrfs/tree-log.h +3 -0
index 1620f8170629..57ab5f3b8dc7 100644
@@ -12,6 +12,9 @@
 /* return value for btrfs_log_dentry_safe that means we don't need to log it at all */
 #define BTRFS_NO_LOG_SYNC 256
 
+/* We can't use the tree log for whatever reason, force a transaction commit */
+#define BTRFS_LOG_FORCE_COMMIT				(1)
+
 struct btrfs_log_ctx {
 	int log_ret;
 	int log_transid;
Signature · 1 line
2.35.1
author maintainer
#1015 lore raw

PATCH 1013/1095 btrfs: ensure pages are unlocked on cow_file_range() failure

In reply to Greg Kroah-Hartman

From: Naohiro Aota <[email protected]>

[ Upstream commit 9ce7466f372d83054c7494f6b3e4b9abaf3f0355 ]

There is a hung_task report on zoned btrfs like below.

https://github.com/naota/linux/issues/59

  [726.328648] INFO: task rocksdb:high0:11085 blocked for more than 241 seconds.
  [726.329839]       Not tainted 5.16.0-rc1+ #1
  [726.330484] "echo 0 > /proc/sys/kernel/hung_task_timeout_secs" disables this message.
  [726.331603] task:rocksdb:high0   state:D stack:    0 pid:11085 ppid: 11082 flags:0x00000000
  [726.331608] Call Trace:
  [726.331611]  <TASK>
  [726.331614]  __schedule+0x2e5/0x9d0
  [726.331622]  schedule+0x58/0xd0
  [726.331626]  io_schedule+0x3f/0x70
  [726.331629]  __folio_lock+0x125/0x200
  [726.331634]  ? find_get_entries+0x1bc/0x240
  [726.331638]  ? filemap_invalidate_unlock_two+0x40/0x40
  [726.331642]  truncate_inode_pages_range+0x5b2/0x770
  [726.331649]  truncate_inode_pages_final+0x44/0x50
  [726.331653]  btrfs_evict_inode+0x67/0x480
  [726.331658]  evict+0xd0/0x180
  [726.331661]  iput+0x13f/0x200
  [726.331664]  do_unlinkat+0x1c0/0x2b0
  [726.331668]  __x64_sys_unlink+0x23/0x30
  [726.331670]  do_syscall_64+0x3b/0xc0
  [726.331674]  entry_SYSCALL_64_after_hwframe+0x44/0xae
  [726.331677] RIP: 0033:0x7fb9490a171b
  [726.331681] RSP: 002b:00007fb943ffac68 EFLAGS: 00000246 ORIG_RAX: 0000000000000057
  [726.331684] RAX: ffffffffffffffda RBX: 0000000000000000 RCX: 00007fb9490a171b
  [726.331686] RDX: 00007fb943ffb040 RSI: 000055a6bbe6ec20 RDI: 00007fb94400d300
  [726.331687] RBP: 00007fb943ffad00 R08: 0000000000000000 R09: 0000000000000000
  [726.331688] R10: 0000000000000031 R11: 0000000000000246 R12: 00007fb943ffb000
  [726.331690] R13: 00007fb943ffb040 R14: 0000000000000000 R15: 00007fb943ffd260
  [726.331693]  </TASK>

While we debug the issue, we found running fstests generic/551 on 5GB non-zoned null_blk device in the emulated zoned mode also had a similar hung issue.

Also, we can reproduce the same symptom with an error injected cow_file_range() setup.

The hang occurs when cow_file_range() fails in the middle of allocation. cow_file_range() called from do_allocation_zoned() can split the give region ([start, end]) for allocation depending on current block group usages. When btrfs can allocate bytes for one part of the split regions but fails for the other region (e.g. because of -ENOSPC), we return the error leaving the pages in the succeeded regions locked. Technically, this occurs only when @unlock == 0. Otherwise, we unlock the pages in an allocated region after creating an ordered extent.

Considering the callers of cow_file_range(unlock=0) won't write out the pages, we can unlock the pages on error exit from cow_file_range(). So, we can ensure all the pages except @locked_page are unlocked on error case.

In summary, cow_file_range now behaves like this:

- page_started == 1 (return value)
- All the pages are unlocked. IO is started.
- unlock == 1
- All the pages except @locked_page are unlocked in any case
- unlock == 0
- On success, all the pages are locked for writing out them
- On failure, all the pages except @locked_page are unlocked

Fixes 42c011000963 ("btrfs: zoned: introduce dedicated data write path for zoned filesystems")
---
 fs/btrfs/inode.c | 72 ++++++++++++++++++++++++++++++++++++++++++------
 1 file changed, 64 insertions(+), 8 deletions(-)
fs/btrfs/inode.c +64 -8
index 5d15e374d032..54afa9e538c5 100644
@@ -1097,6 +1097,28 @@ static u64 get_extent_allocation_hint(struct btrfs_inode *inode, u64 start,
  * *page_started is set to one if we unlock locked_page and do everything
  * required to start IO on it.  It may be clean and already done with
  * IO when we return.
+ *
+ * When unlock == 1, we unlock the pages in successfully allocated regions.
+ * When unlock == 0, we leave them locked for writing them out.
+ *
+ * However, we unlock all the pages except @locked_page in case of failure.
+ *
+ * In summary, page locking state will be as follow:
+ *
+ * - page_started == 1 (return value)
+ *     - All the pages are unlocked. IO is started.
+ *     - Note that this can happen only on success
+ * - unlock == 1
+ *     - All the pages except @locked_page are unlocked in any case
+ * - unlock == 0
+ *     - On success, all the pages are locked for writing out them
+ *     - On failure, all the pages except @locked_page are unlocked
+ *
+ * When a failure happens in the second or later iteration of the
+ * while-loop, the ordered extents created in previous iterations are kept
+ * intact. So, the caller must clean them up by calling
+ * btrfs_cleanup_ordered_extents(). See btrfs_run_delalloc_range() for
+ * example.
  */
 static noinline int cow_file_range(struct btrfs_inode *inode,
 				   struct page *locked_page,
@@ -1106,6 +1128,7 @@ static noinline int cow_file_range(struct btrfs_inode *inode,
 	struct btrfs_root *root = inode->root;
 	struct btrfs_fs_info *fs_info = root->fs_info;
 	u64 alloc_hint = 0;
+	u64 orig_start = start;
 	u64 num_bytes;
 	unsigned long ram_size;
 	u64 cur_alloc_size = 0;
@@ -1293,18 +1316,44 @@ static noinline int cow_file_range(struct btrfs_inode *inode,
 	btrfs_dec_block_group_reservations(fs_info, ins.objectid);
 	btrfs_free_reserved_extent(fs_info, ins.objectid, ins.offset, 1);
 out_unlock:
+	/*
+	 * Now, we have three regions to clean up:
+	 *
+	 * |-------(1)----|---(2)---|-------------(3)----------|
+	 * `- orig_start  `- start  `- start + cur_alloc_size  `- end
+	 *
+	 * We process each region below.
+	 */
+
 	clear_bits = EXTENT_LOCKED | EXTENT_DELALLOC | EXTENT_DELALLOC_NEW |
 		EXTENT_DEFRAG | EXTENT_CLEAR_META_RESV;
 	page_ops = PAGE_UNLOCK | PAGE_START_WRITEBACK | PAGE_END_WRITEBACK;
+
 	/*
-	 * If we reserved an extent for our delalloc range (or a subrange) and
-	 * failed to create the respective ordered extent, then it means that
-	 * when we reserved the extent we decremented the extent's size from
-	 * the data space_info's bytes_may_use counter and incremented the
-	 * space_info's bytes_reserved counter by the same amount. We must make
-	 * sure extent_clear_unlock_delalloc() does not try to decrement again
-	 * the data space_info's bytes_may_use counter, therefore we do not pass
-	 * it the flag EXTENT_CLEAR_DATA_RESV.
+	 * For the range (1). We have already instantiated the ordered extents
+	 * for this region. They are cleaned up by
+	 * btrfs_cleanup_ordered_extents() in e.g,
+	 * btrfs_run_delalloc_range(). EXTENT_LOCKED | EXTENT_DELALLOC are
+	 * already cleared in the above loop. And, EXTENT_DELALLOC_NEW |
+	 * EXTENT_DEFRAG | EXTENT_CLEAR_META_RESV are handled by the cleanup
+	 * function.
+	 *
+	 * However, in case of unlock == 0, we still need to unlock the pages
+	 * (except @locked_page) to ensure all the pages are unlocked.
+	 */
+	if (!unlock && orig_start < start)
+		extent_clear_unlock_delalloc(inode, orig_start, start - 1,
+					     locked_page, 0, page_ops);
+
+	/*
+	 * For the range (2). If we reserved an extent for our delalloc range
+	 * (or a subrange) and failed to create the respective ordered extent,
+	 * then it means that when we reserved the extent we decremented the
+	 * extent's size from the data space_info's bytes_may_use counter and
+	 * incremented the space_info's bytes_reserved counter by the same
+	 * amount. We must make sure extent_clear_unlock_delalloc() does not try
+	 * to decrement again the data space_info's bytes_may_use counter,
+	 * therefore we do not pass it the flag EXTENT_CLEAR_DATA_RESV.
 	 */
 	if (extent_reserved) {
 		extent_clear_unlock_delalloc(inode, start,
@@ -1316,6 +1365,13 @@ static noinline int cow_file_range(struct btrfs_inode *inode,
 		if (start >= end)
 			goto out;
 	}
+
+	/*
+	 * For the range (3). We never touched the region. In addition to the
+	 * clear_bits above, we add EXTENT_CLEAR_DATA_RESV to release the data
+	 * space_info's bytes_may_use counter, reserved in
+	 * btrfs_check_data_free_space().
+	 */
 	extent_clear_unlock_delalloc(inode, start, end, locked_page,
 				     clear_bits | EXTENT_CLEAR_DATA_RESV,
 				     page_ops);
Signature · 1 line
2.35.1
author maintainer
#1016 lore raw

PATCH 1014/1095 btrfs: fix error handling of fallback uncompress write

In reply to Greg Kroah-Hartman

From: Naohiro Aota <[email protected]>

[ Upstream commit 71aa147b4d9d81fa65afa6016f50d7818b64a54f ]

When cow_file_range() fails in the middle of the allocation loop, it unlocks the pages but leaves the ordered extents intact. Thus, we need to call btrfs_cleanup_ordered_extents() to finish the created ordered extents.

Also, we need to call end_extent_writepage() if locked_page is available because btrfs_cleanup_ordered_extents() never processes the region on the locked_page.

Furthermore, we need to set the mapping as error if locked_page is unavailable before unlocking the pages, so that the errno is properly propagated to the user space.

---
 fs/btrfs/inode.c | 17 +++++++++++++++--
 1 file changed, 15 insertions(+), 2 deletions(-)
fs/btrfs/inode.c +15 -2
index 54afa9e538c5..1e404476fe6a 100644
@@ -891,8 +891,18 @@ static int submit_uncompressed_range(struct btrfs_inode *inode,
 		goto out;
 	}
 	if (ret < 0) {
-		if (locked_page)
+		btrfs_cleanup_ordered_extents(inode, locked_page, start, end - start + 1);
+		if (locked_page) {
+			const u64 page_start = page_offset(locked_page);
+			const u64 page_end = page_start + PAGE_SIZE - 1;
+
+			btrfs_page_set_error(inode->root->fs_info, locked_page,
+					     page_start, PAGE_SIZE);
+			set_page_writeback(locked_page);
+			end_page_writeback(locked_page);
+			end_extent_writepage(locked_page, ret, page_start, page_end);
 			unlock_page(locked_page);
+		}
 		goto out;
 	}
 
@@ -1341,9 +1351,12 @@ static noinline int cow_file_range(struct btrfs_inode *inode,
 	 * However, in case of unlock == 0, we still need to unlock the pages
 	 * (except @locked_page) to ensure all the pages are unlocked.
 	 */
-	if (!unlock && orig_start < start)
+	if (!unlock && orig_start < start) {
+		if (!locked_page)
+			mapping_set_error(inode->vfs_inode.i_mapping, ret);
 		extent_clear_unlock_delalloc(inode, orig_start, start - 1,
 					     locked_page, 0, page_ops);
+	}
 
 	/*
 	 * For the range (2). If we reserved an extent for our delalloc range
Signature · 1 line
2.35.1
author maintainer
#1017 lore raw

PATCH 1015/1095 btrfs: reset block group chunk force if we have to wait

In reply to Greg Kroah-Hartman

From: Josef Bacik <[email protected]>

[ Upstream commit 1314ca78b2c35d3e7d0f097268a2ee6dc0d369ef ]

If you try to force a chunk allocation, but you race with another chunk allocation, you will end up waiting on the chunk allocation that just occurred and then allocate another chunk. If you have many threads all doing this at once you can way over-allocate chunks.

Fix this by resetting force to NO_FORCE, that way if we think we need to allocate we can, otherwise we don't force another chunk allocation if one is already happening.

---
 fs/btrfs/block-group.c | 1 +
 1 file changed, 1 insertion(+)
fs/btrfs/block-group.c +1 -0
index 667b7025d503..1deca5164c23 100644
@@ -3724,6 +3724,7 @@ int btrfs_chunk_alloc(struct btrfs_trans_handle *trans, u64 flags,
 			 * attempt.
 			 */
 			wait_for_alloc = true;
+			force = CHUNK_ALLOC_NO_FORCE;
 			spin_unlock(&space_info->lock);
 			mutex_lock(&fs_info->chunk_mutex);
 			mutex_unlock(&fs_info->chunk_mutex);
Signature · 1 line
2.35.1
author maintainer
#1018 lore raw

PATCH 1016/1095 btrfs: properly flag filesystem with BTRFS_FEATURE_INCOMPAT_BIG_METADATA

In reply to Greg Kroah-Hartman

From: Nikolay Borisov <[email protected]>

[ Upstream commit e26b04c4c91925dba57324db177a24e18e2d0013 ]

Commit 6f93e834fa7c seemingly inadvertently moved the code responsible for flagging the filesystem as having BIG_METADATA to a place where setting the flag was essentially lost. This means that filesystems created with kernels containing this bug (starting with 5.15) can potentially be mounted by older (pre-3.4) kernels. In reality chances for this happening are low because there are other incompat flags introduced in the mean time. Still the correct behavior is to set INCOMPAT_BIG_METADATA flag and persist this in the superblock.

Fixes 6f93e834fa7c ("btrfs: fix upper limit for max_inline for page size 64K")
---
 fs/btrfs/disk-io.c | 21 +++++++++++----------
 1 file changed, 11 insertions(+), 10 deletions(-)
fs/btrfs/disk-io.c +11 -10
index f45470798022..34cd57d799e4 100644
@@ -3577,16 +3577,6 @@ int __cold open_ctree(struct super_block *sb, struct btrfs_fs_devices *fs_device
 	 */
 	fs_info->compress_type = BTRFS_COMPRESS_ZLIB;
 
-	/*
-	 * Flag our filesystem as having big metadata blocks if they are bigger
-	 * than the page size.
-	 */
-	if (btrfs_super_nodesize(disk_super) > PAGE_SIZE) {
-		if (!(features & BTRFS_FEATURE_INCOMPAT_BIG_METADATA))
-			btrfs_info(fs_info,
-				"flagging fs with big metadata feature");
-		features |= BTRFS_FEATURE_INCOMPAT_BIG_METADATA;
-	}
 
 	/* Set up fs_info before parsing mount options */
 	nodesize = btrfs_super_nodesize(disk_super);
@@ -3627,6 +3617,17 @@ int __cold open_ctree(struct super_block *sb, struct btrfs_fs_devices *fs_device
 	if (features & BTRFS_FEATURE_INCOMPAT_SKINNY_METADATA)
 		btrfs_info(fs_info, "has skinny extents");
 
+	/*
+	 * Flag our filesystem as having big metadata blocks if they are bigger
+	 * than the page size.
+	 */
+	if (btrfs_super_nodesize(disk_super) > PAGE_SIZE) {
+		if (!(features & BTRFS_FEATURE_INCOMPAT_BIG_METADATA))
+			btrfs_info(fs_info,
+				"flagging fs with big metadata feature");
+		features |= BTRFS_FEATURE_INCOMPAT_BIG_METADATA;
+	}
+
 	/*
 	 * mixed block groups end up with duplicate but slightly offset
 	 * extent buffers for the same range.  It leads to corruptions
Signature · 1 line
2.35.1
author maintainer
#1019 lore raw

PATCH 1017/1095 block: add a bdev_max_zone_append_sectors helper

In reply to Greg Kroah-Hartman

From: Christoph Hellwig <[email protected]>

[ Upstream commit 2aba0d19f4d8c8929b4b3b94a9cfde2aa20e6ee2 ]

Add a helper to check the max supported sectors for zone append based on the block_device instead of having to poke into the block layer internal request_queue.

---
 drivers/nvme/target/zns.c | 3 +--
 fs/zonefs/super.c         | 3 +--
 include/linux/blkdev.h    | 6 ++++++
 3 files changed, 8 insertions(+), 4 deletions(-)
drivers/nvme/target/zns.c +1 -2
index e34718b09550..82b61acf7a72 100644
@@ -34,8 +34,7 @@ static int validate_conv_zones_cb(struct blk_zone *z,
 
 bool nvmet_bdev_zns_enable(struct nvmet_ns *ns)
 {
-	struct request_queue *q = ns->bdev->bd_disk->queue;
-	u8 zasl = nvmet_zasl(queue_max_zone_append_sectors(q));
+	u8 zasl = nvmet_zasl(bdev_max_zone_append_sectors(ns->bdev));
 	struct gendisk *bd_disk = ns->bdev->bd_disk;
 	int ret;
 
fs/zonefs/super.c +1 -2
index 15a4c7c07a3b..b68798a572fc 100644
@@ -723,13 +723,12 @@ static ssize_t zonefs_file_dio_append(struct kiocb *iocb, struct iov_iter *from)
 	struct inode *inode = file_inode(iocb->ki_filp);
 	struct zonefs_inode_info *zi = ZONEFS_I(inode);
 	struct block_device *bdev = inode->i_sb->s_bdev;
-	unsigned int max;
+	unsigned int max = bdev_max_zone_append_sectors(bdev);
 	struct bio *bio;
 	ssize_t size;
 	int nr_pages;
 	ssize_t ret;
 
-	max = queue_max_zone_append_sectors(bdev_get_queue(bdev));
 	max = ALIGN_DOWN(max << SECTOR_SHIFT, inode->i_sb->s_blocksize);
 	iov_iter_truncate(from, max);
 
include/linux/blkdev.h +6 -0
index cc6b24a5098f..34f2b88dfd6e 100644
@@ -1190,6 +1190,12 @@ static inline unsigned int queue_max_zone_append_sectors(const struct request_qu
 	return min(l->max_zone_append_sectors, l->max_sectors);
 }
 
+static inline unsigned int
+bdev_max_zone_append_sectors(struct block_device *bdev)
+{
+	return queue_max_zone_append_sectors(bdev_get_queue(bdev));
+}
+
 static inline unsigned queue_logical_block_size(const struct request_queue *q)
 {
 	int retval = 512;
Signature · 1 line
2.35.1
author maintainer
#1020 lore raw

PATCH 1018/1095 block: add bdev_max_segments() helper

In reply to Greg Kroah-Hartman

From: Naohiro Aota <[email protected]>

[ Upstream commit 65ea1b66482f415d51cd46515b02477257330339 ]

Add bdev_max_segments() like other queue parameters.

---
 include/linux/blkdev.h | 5 +++++
 1 file changed, 5 insertions(+)
include/linux/blkdev.h +5 -0
index 34f2b88dfd6e..7927480b9cf7 100644
@@ -1196,6 +1196,11 @@ bdev_max_zone_append_sectors(struct block_device *bdev)
 	return queue_max_zone_append_sectors(bdev_get_queue(bdev));
 }
 
+static inline unsigned int bdev_max_segments(struct block_device *bdev)
+{
+	return queue_max_segments(bdev_get_queue(bdev));
+}
+
 static inline unsigned queue_logical_block_size(const struct request_queue *q)
 {
 	int retval = 512;
Signature · 1 line
2.35.1
author maintainer
#1021 lore raw

PATCH 1019/1095 btrfs: zoned: revive max_zone_append_bytes

In reply to Greg Kroah-Hartman

From: Naohiro Aota <[email protected]>

[ Upstream commit c2ae7b772ef4e86c5ddf3fd47bf59045ae96a414 ]

This patch is basically a revert of commit 5a80d1c6a270 ("btrfs: zoned:
remove max_zone_append_size logic"), but without unnecessary ASSERT and check. The max_zone_append_size will be used as a hint to estimate the number of extents to cover delalloc/writeback region in the later commits.

The size of a ZONE APPEND bio is also limited by queue_max_segments(), so this commit considers it to calculate max_zone_append_size. Technically, a bio can be larger than queue_max_segments() * PAGE_SIZE if the pages are contiguous. But, it is safe to consider "queue_max_segments() * PAGE_SIZE" as an upper limit of an extent size to calculate the number of extents needed to write data.

---
 fs/btrfs/ctree.h |  2 ++
 fs/btrfs/zoned.c | 17 +++++++++++++++++
 fs/btrfs/zoned.h |  1 +
 3 files changed, 20 insertions(+)
fs/btrfs/ctree.h +2 -0
index 077c95e9baa5..1c377bcfe787 100644
@@ -1050,6 +1050,8 @@ struct btrfs_fs_info {
 		u64 zoned;
 	};
 
+	/* Max size to emit ZONE_APPEND write command */
+	u64 max_zone_append_size;
 	struct mutex zoned_meta_io_lock;
 	spinlock_t treelog_bg_lock;
 	u64 treelog_bg;
fs/btrfs/zoned.c +17 -0
index 84b6d39509bd..1d5b9308f5ef 100644
@@ -407,6 +407,16 @@ int btrfs_get_dev_zone_info(struct btrfs_device *device, bool populate_cache)
 	nr_sectors = bdev_nr_sectors(bdev);
 	zone_info->zone_size_shift = ilog2(zone_info->zone_size);
 	zone_info->nr_zones = nr_sectors >> ilog2(zone_sectors);
+	/*
+	 * We limit max_zone_append_size also by max_segments *
+	 * PAGE_SIZE. Technically, we can have multiple pages per segment. But,
+	 * since btrfs adds the pages one by one to a bio, and btrfs cannot
+	 * increase the metadata reservation even if it increases the number of
+	 * extents, it is safe to stick with the limit.
+	 */
+	zone_info->max_zone_append_size =
+		min_t(u64, (u64)bdev_max_zone_append_sectors(bdev) << SECTOR_SHIFT,
+		      (u64)bdev_max_segments(bdev) << PAGE_SHIFT);
 	if (!IS_ALIGNED(nr_sectors, zone_sectors))
 		zone_info->nr_zones++;
 
@@ -632,6 +642,7 @@ int btrfs_check_zoned_mode(struct btrfs_fs_info *fs_info)
 	u64 zoned_devices = 0;
 	u64 nr_devices = 0;
 	u64 zone_size = 0;
+	u64 max_zone_append_size = 0;
 	const bool incompat_zoned = btrfs_fs_incompat(fs_info, ZONED);
 	int ret = 0;
 
@@ -666,6 +677,11 @@ int btrfs_check_zoned_mode(struct btrfs_fs_info *fs_info)
 				ret = -EINVAL;
 				goto out;
 			}
+			if (!max_zone_append_size ||
+			    (zone_info->max_zone_append_size &&
+			     zone_info->max_zone_append_size < max_zone_append_size))
+				max_zone_append_size =
+					zone_info->max_zone_append_size;
 		}
 		nr_devices++;
 	}
@@ -715,6 +731,7 @@ int btrfs_check_zoned_mode(struct btrfs_fs_info *fs_info)
 	}
 
 	fs_info->zone_size = zone_size;
+	fs_info->max_zone_append_size = max_zone_append_size;
 	fs_info->fs_devices->chunk_alloc_policy = BTRFS_CHUNK_ALLOC_ZONED;
 
 	/*
fs/btrfs/zoned.h +1 -0
index cf6320feef46..2d6da8f4b55a 100644
@@ -23,6 +23,7 @@ struct btrfs_zoned_device_info {
 	 */
 	u64 zone_size;
 	u8  zone_size_shift;
+	u64 max_zone_append_size;
 	u32 nr_zones;
 	unsigned int max_active_zones;
 	atomic_t active_zones_left;
Signature · 1 line
2.35.1
author maintainer
#1022 lore raw

PATCH 1020/1095 btrfs: replace BTRFS_MAX_EXTENT_SIZE with fs_info->max_extent_size

In reply to Greg Kroah-Hartman

From: Naohiro Aota <[email protected]>

[ Upstream commit f7b12a62f008a3041f42f2426983e59a6a0a3c59 ]

On zoned filesystem, data write out is limited by max_zone_append_size, and a large ordered extent is split according the size of a bio. OTOH, the number of extents to be written is calculated using BTRFS_MAX_EXTENT_SIZE, and that estimated number is used to reserve the metadata bytes to update and/or create the metadata items.

The metadata reservation is done at e.g, btrfs_buffered_write() and then released according to the estimation changes. Thus, if the number of extent increases massively, the reserved metadata can run out.

The increase of the number of extents easily occurs on zoned filesystem if BTRFS_MAX_EXTENT_SIZE > max_zone_append_size. And, it causes the following warning on a small RAM environment with disabling metadata over-commit (in the following patch).

[75721.498492] ------------[ cut here ]------------
[75721.505624] BTRFS: block rsv 1 returned -28
[75721.512230] WARNING: CPU: 24 PID: 2327559 at fs/btrfs/block-rsv.c:537 btrfs_use_block_rsv+0x560/0x760 [btrfs]
[75721.581854] CPU: 24 PID: 2327559 Comm: kworker/u64:10 Kdump: loaded Tainted: G        W         5.18.0-rc2-BTRFS-ZNS+ #109
[75721.597200] Hardware name: Supermicro Super Server/H12SSL-NT, BIOS 2.0 02/22/2021
[75721.607310] Workqueue: btrfs-endio-write btrfs_work_helper [btrfs]
[75721.616209] RIP: 0010:btrfs_use_block_rsv+0x560/0x760 [btrfs]
[75721.646649] RSP: 0018:ffffc9000fbdf3e0 EFLAGS: 00010286
[75721.654126] RAX: 0000000000000000 RBX: 0000000000004000 RCX: 0000000000000000
[75721.663524] RDX: 0000000000000004 RSI: 0000000000000008 RDI: fffff52001f7be6e
[75721.672921] RBP: ffffc9000fbdf420 R08: 0000000000000001 R09: ffff889f8d1fc6c7
[75721.682493] R10: ffffed13f1a3f8d8 R11: 0000000000000001 R12: ffff88980a3c0e28
[75721.692284] R13: ffff889b66590000 R14: ffff88980a3c0e40 R15: ffff88980a3c0e8a
[75721.701878] FS:  0000000000000000(0000) GS:ffff889f8d000000(0000) knlGS:0000000000000000
[75721.712601] CS:  0010 DS: 0000 ES: 0000 CR0: 0000000080050033
[75721.720726] CR2: 000055d12e05c018 CR3: 0000800193594000 CR4: 0000000000350ee0
[75721.730499] Call Trace:
[75721.735166]  <TASK>
[75721.739886]  btrfs_alloc_tree_block+0x1e1/0x1100 [btrfs]
[75721.747545]  ? btrfs_alloc_logged_file_extent+0x550/0x550 [btrfs]
[75721.756145]  ? btrfs_get_32+0xea/0x2d0 [btrfs]
[75721.762852]  ? btrfs_get_32+0xea/0x2d0 [btrfs]
[75721.769520]  ? push_leaf_left+0x420/0x620 [btrfs]
[75721.776431]  ? memcpy+0x4e/0x60
[75721.781931]  split_leaf+0x433/0x12d0 [btrfs]
[75721.788392]  ? btrfs_get_token_32+0x580/0x580 [btrfs]
[75721.795636]  ? push_for_double_split.isra.0+0x420/0x420 [btrfs]
[75721.803759]  ? leaf_space_used+0x15d/0x1a0 [btrfs]
[75721.811156]  btrfs_search_slot+0x1bc3/0x2790 [btrfs]
[75721.818300]  ? lock_downgrade+0x7c0/0x7c0
[75721.824411]  ? free_extent_buffer.part.0+0x107/0x200 [btrfs]
[75721.832456]  ? split_leaf+0x12d0/0x12d0 [btrfs]
[75721.839149]  ? free_extent_buffer.part.0+0x14f/0x200 [btrfs]
[75721.846945]  ? free_extent_buffer+0x13/0x20 [btrfs]
[75721.853960]  ? btrfs_release_path+0x4b/0x190 [btrfs]
[75721.861429]  btrfs_csum_file_blocks+0x85c/0x1500 [btrfs]
[75721.869313]  ? rcu_read_lock_sched_held+0x16/0x80
[75721.876085]  ? lock_release+0x552/0xf80
[75721.881957]  ? btrfs_del_csums+0x8c0/0x8c0 [btrfs]
[75721.888886]  ? __kasan_check_write+0x14/0x20
[75721.895152]  ? do_raw_read_unlock+0x44/0x80
[75721.901323]  ? _raw_write_lock_irq+0x60/0x80
[75721.907983]  ? btrfs_global_root+0xb9/0xe0 [btrfs]
[75721.915166]  ? btrfs_csum_root+0x12b/0x180 [btrfs]
[75721.921918]  ? btrfs_get_global_root+0x820/0x820 [btrfs]
[75721.929166]  ? _raw_write_unlock+0x23/0x40
[75721.935116]  ? unpin_extent_cache+0x1e3/0x390 [btrfs]
[75721.942041]  btrfs_finish_ordered_io.isra.0+0xa0c/0x1dc0 [btrfs]
[75721.949906]  ? try_to_wake_up+0x30/0x14a0
[75721.955700]  ? btrfs_unlink_subvol+0xda0/0xda0 [btrfs]
[75721.962661]  ? rcu_read_lock_sched_held+0x16/0x80
[75721.969111]  ? lock_acquire+0x41b/0x4c0
[75721.974982]  finish_ordered_fn+0x15/0x20 [btrfs]
[75721.981639]  btrfs_work_helper+0x1af/0xa80 [btrfs]
[75721.988184]  ? _raw_spin_unlock_irq+0x28/0x50
[75721.994643]  process_one_work+0x815/0x1460
[75722.000444]  ? pwq_dec_nr_in_flight+0x250/0x250
[75722.006643]  ? do_raw_spin_trylock+0xbb/0x190
[75722.013086]  worker_thread+0x59a/0xeb0
[75722.018511]  kthread+0x2ac/0x360
[75722.023428]  ? process_one_work+0x1460/0x1460
[75722.029431]  ? kthread_complete_and_exit+0x30/0x30
[75722.036044]  ret_from_fork+0x22/0x30
[75722.041255]  </TASK>
[75722.045047] irq event stamp: 0
[75722.049703] hardirqs last  enabled at (0): [<0000000000000000>] 0x0
[75722.057610] hardirqs last disabled at (0): [<ffffffff8118a94a>] copy_process+0x1c1a/0x66b0
[75722.067533] softirqs last  enabled at (0): [<ffffffff8118a989>] copy_process+0x1c59/0x66b0
[75722.077423] softirqs last disabled at (0): [<0000000000000000>] 0x0
[75722.085335] ---[ end trace 0000000000000000 ]---

To fix the estimation, we need to introduce fs_info->max_extent_size to replace BTRFS_MAX_EXTENT_SIZE, which allow setting the different size for regular vs zoned filesystem.

Set fs_info->max_extent_size to BTRFS_MAX_EXTENT_SIZE by default. On zoned filesystem, it is set to fs_info->max_zone_append_size.

Fixes d8e3fb106f39 ("btrfs: zoned: use ZONE_APPEND write for zoned mode")
---
 fs/btrfs/ctree.h     | 6 ++++++
 fs/btrfs/disk-io.c   | 2 ++
 fs/btrfs/extent_io.c | 4 +++-
 fs/btrfs/inode.c     | 6 ++++--
 fs/btrfs/zoned.c     | 5 ++++-
 5 files changed, 19 insertions(+), 4 deletions(-)
fs/btrfs/ctree.h +6 -0
index 1c377bcfe787..97f5a3d320ff 100644
@@ -1032,6 +1032,12 @@ struct btrfs_fs_info {
 	u32 csums_per_leaf;
 	u32 stripesize;
 
+	/*
+	 * Maximum size of an extent. BTRFS_MAX_EXTENT_SIZE on regular
+	 * filesystem, on zoned it depends on the device constraints.
+	 */
+	u64 max_extent_size;
+
 	/* Block groups and devices containing active swapfiles. */
 	spinlock_t swapfile_pins_lock;
 	struct rb_root swapfile_pins;
fs/btrfs/disk-io.c +2 -0
index 34cd57d799e4..bf5c6ac67e87 100644
@@ -3246,6 +3246,8 @@ void btrfs_init_fs_info(struct btrfs_fs_info *fs_info)
 	fs_info->sectorsize_bits = ilog2(4096);
 	fs_info->stripesize = 4096;
 
+	fs_info->max_extent_size = BTRFS_MAX_EXTENT_SIZE;
+
 	spin_lock_init(&fs_info->swapfile_pins_lock);
 	fs_info->swapfile_pins = RB_ROOT;
 
fs/btrfs/extent_io.c +3 -1
index 68ddd90685d9..bfc7d5b31156 100644
@@ -1992,10 +1992,12 @@ noinline_for_stack bool find_lock_delalloc_range(struct inode *inode,
 				    struct page *locked_page, u64 *start,
 				    u64 *end)
 {
+	struct btrfs_fs_info *fs_info = btrfs_sb(inode->i_sb);
 	struct extent_io_tree *tree = &BTRFS_I(inode)->io_tree;
 	const u64 orig_start = *start;
 	const u64 orig_end = *end;
-	u64 max_bytes = BTRFS_MAX_EXTENT_SIZE;
+	/* The sanity tests may not set a valid fs_info. */
+	u64 max_bytes = fs_info ? fs_info->max_extent_size : BTRFS_MAX_EXTENT_SIZE;
 	u64 delalloc_start;
 	u64 delalloc_end;
 	bool found;
fs/btrfs/inode.c +4 -2
index 1e404476fe6a..c50288d90c66 100644
@@ -2102,6 +2102,7 @@ int btrfs_run_delalloc_range(struct btrfs_inode *inode, struct page *locked_page
 void btrfs_split_delalloc_extent(struct inode *inode,
 				 struct extent_state *orig, u64 split)
 {
+	struct btrfs_fs_info *fs_info = btrfs_sb(inode->i_sb);
 	u64 size;
 
 	/* not delalloc, ignore it */
@@ -2109,7 +2110,7 @@ void btrfs_split_delalloc_extent(struct inode *inode,
 		return;
 
 	size = orig->end - orig->start + 1;
-	if (size > BTRFS_MAX_EXTENT_SIZE) {
+	if (size > fs_info->max_extent_size) {
 		u32 num_extents;
 		u64 new_size;
 
@@ -2138,6 +2139,7 @@ void btrfs_split_delalloc_extent(struct inode *inode,
 void btrfs_merge_delalloc_extent(struct inode *inode, struct extent_state *new,
 				 struct extent_state *other)
 {
+	struct btrfs_fs_info *fs_info = btrfs_sb(inode->i_sb);
 	u64 new_size, old_size;
 	u32 num_extents;
 
@@ -2151,7 +2153,7 @@ void btrfs_merge_delalloc_extent(struct inode *inode, struct extent_state *new,
 		new_size = other->end - new->start + 1;
 
 	/* we're not bigger than the max, unreserve the space and go */
-	if (new_size <= BTRFS_MAX_EXTENT_SIZE) {
+	if (new_size <= fs_info->max_extent_size) {
 		spin_lock(&BTRFS_I(inode)->lock);
 		btrfs_mod_outstanding_extents(BTRFS_I(inode), -1);
 		spin_unlock(&BTRFS_I(inode)->lock);
fs/btrfs/zoned.c +4 -1
index 1d5b9308f5ef..a0bf2c20fa61 100644
@@ -731,8 +731,11 @@ int btrfs_check_zoned_mode(struct btrfs_fs_info *fs_info)
 	}
 
 	fs_info->zone_size = zone_size;
-	fs_info->max_zone_append_size = max_zone_append_size;
+	fs_info->max_zone_append_size = ALIGN_DOWN(max_zone_append_size,
+						   fs_info->sectorsize);
 	fs_info->fs_devices->chunk_alloc_policy = BTRFS_CHUNK_ALLOC_ZONED;
+	if (fs_info->max_zone_append_size < fs_info->max_extent_size)
+		fs_info->max_extent_size = fs_info->max_zone_append_size;
 
 	/*
 	 * Check mount options here, because we might change fs_info->zoned
Signature · 1 line
2.35.1
author maintainer
#1023 lore raw

PATCH 1021/1095 btrfs: let can_allocate_chunk return error

In reply to Greg Kroah-Hartman

From: Naohiro Aota <[email protected]>

[ Upstream commit bb9950d3df7169a673c594d38fb74e241ed4fb2a ]

For the later patch, convert the return type from bool to int and return errors. No functional changes.

---
 fs/btrfs/extent-tree.c | 15 ++++++++-------
 1 file changed, 8 insertions(+), 7 deletions(-)
fs/btrfs/extent-tree.c +8 -7
index f45ecd939a2c..8bdcbc0c6d60 100644
@@ -3985,12 +3985,12 @@ static void found_extent(struct find_free_extent_ctl *ffe_ctl,
 	}
 }
 
-static bool can_allocate_chunk(struct btrfs_fs_info *fs_info,
-			       struct find_free_extent_ctl *ffe_ctl)
+static int can_allocate_chunk(struct btrfs_fs_info *fs_info,
+			      struct find_free_extent_ctl *ffe_ctl)
 {
 	switch (ffe_ctl->policy) {
 	case BTRFS_EXTENT_ALLOC_CLUSTERED:
-		return true;
+		return 0;
 	case BTRFS_EXTENT_ALLOC_ZONED:
 		/*
 		 * If we have enough free space left in an already
@@ -4000,8 +4000,8 @@ static bool can_allocate_chunk(struct btrfs_fs_info *fs_info,
 		 */
 		if (ffe_ctl->max_extent_size >= ffe_ctl->min_alloc_size &&
 		    !btrfs_can_activate_zone(fs_info->fs_devices, ffe_ctl->flags))
-			return false;
-		return true;
+			return -ENOSPC;
+		return 0;
 	default:
 		BUG();
 	}
@@ -4083,8 +4083,9 @@ static int find_free_extent_update_loop(struct btrfs_fs_info *fs_info,
 			int exist = 0;
 
 			/*Check if allocation policy allows to create a new chunk */
-			if (!can_allocate_chunk(fs_info, ffe_ctl))
-				return -ENOSPC;
+			ret = can_allocate_chunk(fs_info, ffe_ctl);
+			if (ret)
+				return ret;
 
 			trans = current->journal_info;
 			if (trans)
Signature · 1 line
2.35.1
author maintainer
#1024 lore raw

PATCH 1022/1095 btrfs: zoned: finish least available block group on data bg allocation

In reply to Greg Kroah-Hartman

From: Naohiro Aota <[email protected]>

[ Upstream commit 393f646e34c18b85d0f41272bfcbd475ae3a0d34 ]

When we run out of active zones and no sufficient space is left in any block groups, we need to finish one block group to make room to activate a new block group.

However, we cannot do this for metadata block groups because we can cause a deadlock by waiting for a running transaction commit. So, do that only for a data block group.

Furthermore, the block group to be finished has two requirements. First, the block group must not have reserved bytes left. Having reserved bytes means we have an allocated region but did not yet send bios for it. If that region is allocated by the thread calling btrfs_zone_finish(), it results in a deadlock.

Second, the block group to be finished must not be a SYSTEM block group. Finishing a SYSTEM block group easily breaks further chunk allocation by nullifying the SYSTEM free space.

In a certain case, we cannot find any zone finish candidate or btrfs_zone_finish() may fail. In that case, we fall back to split the allocation bytes and fill the last spaces left in the block groups.

Fixes afba2bc036b0 ("btrfs: zoned: implement active zone tracking")
---
 fs/btrfs/extent-tree.c | 50 +++++++++++++++++++++++++++++++++---------
 fs/btrfs/zoned.c       | 40 +++++++++++++++++++++++++++++++++
 fs/btrfs/zoned.h       |  7 ++++++
 3 files changed, 87 insertions(+), 10 deletions(-)
fs/btrfs/extent-tree.c +40 -10
index 8bdcbc0c6d60..bdebd77f31b4 100644
@@ -3985,6 +3985,45 @@ static void found_extent(struct find_free_extent_ctl *ffe_ctl,
 	}
 }
 
+static int can_allocate_chunk_zoned(struct btrfs_fs_info *fs_info,
+				    struct find_free_extent_ctl *ffe_ctl)
+{
+	/* If we can activate new zone, just allocate a chunk and use it */
+	if (btrfs_can_activate_zone(fs_info->fs_devices, ffe_ctl->flags))
+		return 0;
+
+	/*
+	 * We already reached the max active zones. Try to finish one block
+	 * group to make a room for a new block group. This is only possible
+	 * for a data block group because btrfs_zone_finish() may need to wait
+	 * for a running transaction which can cause a deadlock for metadata
+	 * allocation.
+	 */
+	if (ffe_ctl->flags & BTRFS_BLOCK_GROUP_DATA) {
+		int ret = btrfs_zone_finish_one_bg(fs_info);
+
+		if (ret == 1)
+			return 0;
+		else if (ret < 0)
+			return ret;
+	}
+
+	/*
+	 * If we have enough free space left in an already active block group
+	 * and we can't activate any other zone now, do not allow allocating a
+	 * new chunk and let find_free_extent() retry with a smaller size.
+	 */
+	if (ffe_ctl->max_extent_size >= ffe_ctl->min_alloc_size)
+		return -ENOSPC;
+
+	/*
+	 * We cannot activate a new block group and no enough space left in any
+	 * block groups. So, allocating a new block group may not help. But,
+	 * there is nothing to do anyway, so let's go with it.
+	 */
+	return 0;
+}
+
 static int can_allocate_chunk(struct btrfs_fs_info *fs_info,
 			      struct find_free_extent_ctl *ffe_ctl)
 {
@@ -3992,16 +4031,7 @@ static int can_allocate_chunk(struct btrfs_fs_info *fs_info,
 	case BTRFS_EXTENT_ALLOC_CLUSTERED:
 		return 0;
 	case BTRFS_EXTENT_ALLOC_ZONED:
-		/*
-		 * If we have enough free space left in an already
-		 * active block group and we can't activate any other
-		 * zone now, do not allow allocating a new chunk and
-		 * let find_free_extent() retry with a smaller size.
-		 */
-		if (ffe_ctl->max_extent_size >= ffe_ctl->min_alloc_size &&
-		    !btrfs_can_activate_zone(fs_info->fs_devices, ffe_ctl->flags))
-			return -ENOSPC;
-		return 0;
+		return can_allocate_chunk_zoned(fs_info, ffe_ctl);
 	default:
 		BUG();
 	}
fs/btrfs/zoned.c +40 -0
index a0bf2c20fa61..0a6a3d6f5af7 100644
@@ -2176,3 +2176,43 @@ void btrfs_zoned_release_data_reloc_bg(struct btrfs_fs_info *fs_info, u64 logica
 	spin_unlock(&block_group->lock);
 	btrfs_put_block_group(block_group);
 }
+
+int btrfs_zone_finish_one_bg(struct btrfs_fs_info *fs_info)
+{
+	struct btrfs_block_group *block_group;
+	struct btrfs_block_group *min_bg = NULL;
+	u64 min_avail = U64_MAX;
+	int ret;
+
+	spin_lock(&fs_info->zone_active_bgs_lock);
+	list_for_each_entry(block_group, &fs_info->zone_active_bgs,
+			    active_bg_list) {
+		u64 avail;
+
+		spin_lock(&block_group->lock);
+		if (block_group->reserved ||
+		    (block_group->flags & BTRFS_BLOCK_GROUP_SYSTEM)) {
+			spin_unlock(&block_group->lock);
+			continue;
+		}
+
+		avail = block_group->zone_capacity - block_group->alloc_offset;
+		if (min_avail > avail) {
+			if (min_bg)
+				btrfs_put_block_group(min_bg);
+			min_bg = block_group;
+			min_avail = avail;
+			btrfs_get_block_group(min_bg);
+		}
+		spin_unlock(&block_group->lock);
+	}
+	spin_unlock(&fs_info->zone_active_bgs_lock);
+
+	if (!min_bg)
+		return 0;
+
+	ret = btrfs_zone_finish(min_bg);
+	btrfs_put_block_group(min_bg);
+
+	return ret < 0 ? ret : 1;
+}
fs/btrfs/zoned.h +7 -0
index 2d6da8f4b55a..c424417e19bb 100644
@@ -83,6 +83,7 @@ void btrfs_clear_data_reloc_bg(struct btrfs_block_group *bg);
 void btrfs_free_zone_cache(struct btrfs_fs_info *fs_info);
 void btrfs_zoned_release_data_reloc_bg(struct btrfs_fs_info *fs_info, u64 logical,
 				       u64 length);
+int btrfs_zone_finish_one_bg(struct btrfs_fs_info *fs_info);
 #else /* CONFIG_BLK_DEV_ZONED */
 static inline int btrfs_get_dev_zone(struct btrfs_device *device, u64 pos,
 				     struct blk_zone *zone)
@@ -247,6 +248,12 @@ static inline void btrfs_free_zone_cache(struct btrfs_fs_info *fs_info) { }
 
 static inline void btrfs_zoned_release_data_reloc_bg(struct btrfs_fs_info *fs_info,
 						     u64 logical, u64 length) { }
+
+static inline int btrfs_zone_finish_one_bg(struct btrfs_fs_info *fs_info)
+{
+	return 1;
+}
+
 #endif
 
 static inline bool btrfs_dev_is_sequential(struct btrfs_device *device, u64 pos)
Signature · 1 line
2.35.1
author maintainer
#1025 lore raw

PATCH 1023/1095 btrfs: zoned: disable metadata overcommit for zoned

In reply to Greg Kroah-Hartman

From: Naohiro Aota <[email protected]>

[ Upstream commit 79417d040f4f77b19c701bccc23013b9cdac358d ]

The metadata overcommit makes the space reservation flexible but it is also harmful to active zone tracking. Since we cannot finish a block group from the metadata allocation context, we might not activate a new block group and might not be able to actually write out the overcommit reservations.

So, disable metadata overcommit for zoned filesystems. We will ensure the reservations are under active_total_bytes in the following patches.

Fixes afba2bc036b0 ("btrfs: zoned: implement active zone tracking")
---
 fs/btrfs/space-info.c | 5 ++++-
 1 file changed, 4 insertions(+), 1 deletion(-)
fs/btrfs/space-info.c +4 -1
index b87931a458eb..56a7c99fc03e 100644
@@ -340,7 +340,10 @@ int btrfs_can_overcommit(struct btrfs_fs_info *fs_info,
 		return 0;
 
 	used = btrfs_space_info_used(space_info, true);
-	avail = calc_available_free_space(fs_info, space_info, flush);
+	if (btrfs_is_zoned(fs_info) && (space_info->flags & BTRFS_BLOCK_GROUP_METADATA))
+		avail = 0;
+	else
+		avail = calc_available_free_space(fs_info, space_info, flush);
 
 	if (used + bytes < space_info->total_bytes + avail)
 		return 1;
Signature · 1 line
2.35.1
author maintainer
#1026 lore raw

PATCH 1024/1095 btrfs: make the bg_reclaim_threshold per-space info

In reply to Greg Kroah-Hartman

From: Josef Bacik <[email protected]>

[ Upstream commit bb5a098d9791f184899499531ff4411089e2a5e0 ]

For non-zoned file systems it's useful to have the auto reclaim feature, however there are different use cases for non-zoned, for example we may not want to reclaim metadata chunks ever, only data chunks. Move this sysfs flag to per-space_info. This won't affect current users because this tunable only ever did anything for zoned, and that is currently hidden behind BTRFS_CONFIG_DEBUG.

---
 fs/btrfs/free-space-cache.c |  7 +++++--
 fs/btrfs/space-info.c       |  9 +++++++++
 fs/btrfs/space-info.h       |  6 ++++++
 fs/btrfs/sysfs.c            | 37 +++++++++++++++++++++++++++++++++++++
 fs/btrfs/zoned.h            |  6 +-----
 5 files changed, 58 insertions(+), 7 deletions(-)
fs/btrfs/free-space-cache.c +5 -2
index 01a408db5683..ef84bc5030cd 100644
@@ -2630,16 +2630,19 @@ int __btrfs_add_free_space(struct btrfs_block_group *block_group,
 static int __btrfs_add_free_space_zoned(struct btrfs_block_group *block_group,
 					u64 bytenr, u64 size, bool used)
 {
-	struct btrfs_fs_info *fs_info = block_group->fs_info;
+	struct btrfs_space_info *sinfo = block_group->space_info;
 	struct btrfs_free_space_ctl *ctl = block_group->free_space_ctl;
 	u64 offset = bytenr - block_group->start;
 	u64 to_free, to_unusable;
-	const int bg_reclaim_threshold = READ_ONCE(fs_info->bg_reclaim_threshold);
+	int bg_reclaim_threshold = 0;
 	bool initial = (size == block_group->length);
 	u64 reclaimable_unusable;
 
 	WARN_ON(!initial && offset + size > block_group->zone_capacity);
 
+	if (!initial)
+		bg_reclaim_threshold = READ_ONCE(sinfo->bg_reclaim_threshold);
+
 	spin_lock(&ctl->tree_lock);
 	if (!used)
 		to_free = size;
fs/btrfs/space-info.c +9 -0
index 56a7c99fc03e..85608acb9557 100644
@@ -181,6 +181,12 @@ void btrfs_clear_space_info_full(struct btrfs_fs_info *info)
 		found->full = 0;
 }
 
+/*
+ * Block groups with more than this value (percents) of unusable space will be
+ * scheduled for background reclaim.
+ */
+#define BTRFS_DEFAULT_ZONED_RECLAIM_THRESH			(75)
+
 static int create_space_info(struct btrfs_fs_info *info, u64 flags)
 {
 
@@ -203,6 +209,9 @@ static int create_space_info(struct btrfs_fs_info *info, u64 flags)
 	INIT_LIST_HEAD(&space_info->priority_tickets);
 	space_info->clamp = 1;
 
+	if (btrfs_is_zoned(info))
+		space_info->bg_reclaim_threshold = BTRFS_DEFAULT_ZONED_RECLAIM_THRESH;
+
 	ret = btrfs_sysfs_add_space_info_type(info, space_info);
 	if (ret)
 		return ret;
fs/btrfs/space-info.h +6 -0
index d841fed73492..a803e29bd781 100644
@@ -24,6 +24,12 @@ struct btrfs_space_info {
 				   the space info if we had an ENOSPC in the
 				   allocator. */
 
+	/*
+	 * Once a block group drops below this threshold (percents) we'll
+	 * schedule it for reclaim.
+	 */
+	int bg_reclaim_threshold;
+
 	int clamp;		/* Used to scale our threshold for preemptive
 				   flushing. The value is >> clamp, so turns
 				   out to be a 2^clamp divisor. */
fs/btrfs/sysfs.c +37 -0
index ba78ca5aabbb..43845cae0c74 100644
@@ -722,6 +722,42 @@ SPACE_INFO_ATTR(bytes_zone_unusable);
 SPACE_INFO_ATTR(disk_used);
 SPACE_INFO_ATTR(disk_total);
 
+static ssize_t btrfs_sinfo_bg_reclaim_threshold_show(struct kobject *kobj,
+						     struct kobj_attribute *a,
+						     char *buf)
+{
+	struct btrfs_space_info *space_info = to_space_info(kobj);
+	ssize_t ret;
+
+	ret = sysfs_emit(buf, "%d\n", READ_ONCE(space_info->bg_reclaim_threshold));
+
+	return ret;
+}
+
+static ssize_t btrfs_sinfo_bg_reclaim_threshold_store(struct kobject *kobj,
+						      struct kobj_attribute *a,
+						      const char *buf, size_t len)
+{
+	struct btrfs_space_info *space_info = to_space_info(kobj);
+	int thresh;
+	int ret;
+
+	ret = kstrtoint(buf, 10, &thresh);
+	if (ret)
+		return ret;
+
+	if (thresh != 0 && (thresh <= 50 || thresh > 100))
+		return -EINVAL;
+
+	WRITE_ONCE(space_info->bg_reclaim_threshold, thresh);
+
+	return len;
+}
+
+BTRFS_ATTR_RW(space_info, bg_reclaim_threshold,
+	      btrfs_sinfo_bg_reclaim_threshold_show,
+	      btrfs_sinfo_bg_reclaim_threshold_store);
+
 /*
  * Allocation information about block group types.
  *
@@ -738,6 +774,7 @@ static struct attribute *space_info_attrs[] = {
 	BTRFS_ATTR_PTR(space_info, bytes_zone_unusable),
 	BTRFS_ATTR_PTR(space_info, disk_used),
 	BTRFS_ATTR_PTR(space_info, disk_total),
+	BTRFS_ATTR_PTR(space_info, bg_reclaim_threshold),
 	NULL,
 };
 ATTRIBUTE_GROUPS(space_info);
fs/btrfs/zoned.h +1 -5
index c424417e19bb..199b69670fa2 100644
@@ -10,11 +10,7 @@
 #include "block-group.h"
 #include "btrfs_inode.h"
 
-/*
- * Block groups with more than this value (percents) of unusable space will be
- * scheduled for background reclaim.
- */
-#define BTRFS_DEFAULT_RECLAIM_THRESH		75
+#define BTRFS_DEFAULT_RECLAIM_THRESH           			(75)
 
 struct btrfs_zoned_device_info {
 	/*
Signature · 1 line
2.35.1
author maintainer
#1027 lore raw

PATCH 1025/1095 btrfs: zoned: introduce btrfs_zoned_bg_is_full

In reply to Greg Kroah-Hartman

From: Naohiro Aota <[email protected]>

[ Upstream commit 1bfd476754a2d63f899ef9c3e253b17766b8fb73 ]

Introduce a wrapper to check if all the space in a block group is allocated or not.

---
 fs/btrfs/extent-tree.c | 3 +--
 fs/btrfs/zoned.c       | 2 +-
 fs/btrfs/zoned.h       | 6 ++++++
 3 files changed, 8 insertions(+), 3 deletions(-)
fs/btrfs/extent-tree.c +1 -2
index bdebd77f31b4..56185541e188 100644
@@ -3803,8 +3803,7 @@ static int do_allocation_zoned(struct btrfs_block_group *block_group,
 
 	/* Check RO and no space case before trying to activate it */
 	spin_lock(&block_group->lock);
-	if (block_group->ro ||
-	    block_group->alloc_offset == block_group->zone_capacity) {
+	if (block_group->ro || btrfs_zoned_bg_is_full(block_group)) {
 		ret = 1;
 		/*
 		 * May need to clear fs_info->{treelog,data_reloc}_bg.
fs/btrfs/zoned.c +1 -1
index 0a6a3d6f5af7..170681797283 100644
@@ -1859,7 +1859,7 @@ bool btrfs_zone_activate(struct btrfs_block_group *block_group)
 	}
 
 	/* No space left */
-	if (block_group->alloc_offset == block_group->zone_capacity) {
+	if (btrfs_zoned_bg_is_full(block_group)) {
 		ret = false;
 		goto out_unlock;
 	}
fs/btrfs/zoned.h +6 -0
index 199b69670fa2..0740458894ac 100644
@@ -384,4 +384,10 @@ static inline void btrfs_zoned_data_reloc_unlock(struct btrfs_inode *inode)
 		mutex_unlock(&root->fs_info->zoned_data_reloc_io_lock);
 }
 
+static inline bool btrfs_zoned_bg_is_full(const struct btrfs_block_group *bg)
+{
+	ASSERT(btrfs_is_zoned(bg->fs_info));
+	return (bg->alloc_offset == bg->zone_capacity);
+}
+
 #endif
Signature · 1 line
2.35.1
author maintainer
#1028 lore raw

PATCH 1026/1095 btrfs: store chunk size in space-info struct

In reply to Greg Kroah-Hartman

From: Stefan Roesch <[email protected]>

[ Upstream commit f6fca3917b4d99d8c13901738afec35f570a3c2f ]

The chunk size is stored in the btrfs_space_info structure. It is initialized at the start and is then used.

A new API is added to update the current chunk size. This API is used to be able to expose the chunk_size as a sysfs setting.

[ rename and merge helpers, switch atomic type to u64, style fixes ]
---
 fs/btrfs/space-info.c | 32 ++++++++++++++++++++++++++++++++
 fs/btrfs/space-info.h |  4 ++++
 fs/btrfs/volumes.c    | 28 +++++++++-------------------
 3 files changed, 45 insertions(+), 19 deletions(-)
fs/btrfs/space-info.c +32 -0
index 85608acb9557..98a84b523be6 100644
@@ -187,6 +187,37 @@ void btrfs_clear_space_info_full(struct btrfs_fs_info *info)
  */
 #define BTRFS_DEFAULT_ZONED_RECLAIM_THRESH			(75)
 
+/*
+ * Calculate chunk size depending on volume type (regular or zoned).
+ */
+static u64 calc_chunk_size(const struct btrfs_fs_info *fs_info, u64 flags)
+{
+	if (btrfs_is_zoned(fs_info))
+		return fs_info->zone_size;
+
+	ASSERT(flags & BTRFS_BLOCK_GROUP_TYPE_MASK);
+
+	if (flags & BTRFS_BLOCK_GROUP_DATA)
+		return SZ_1G;
+	else if (flags & BTRFS_BLOCK_GROUP_SYSTEM)
+		return SZ_32M;
+
+	/* Handle BTRFS_BLOCK_GROUP_METADATA */
+	if (fs_info->fs_devices->total_rw_bytes > 50ULL * SZ_1G)
+		return SZ_1G;
+
+	return SZ_256M;
+}
+
+/*
+ * Update default chunk size.
+ */
+void btrfs_update_space_info_chunk_size(struct btrfs_space_info *space_info,
+					u64 chunk_size)
+{
+	WRITE_ONCE(space_info->chunk_size, chunk_size);
+}
+
 static int create_space_info(struct btrfs_fs_info *info, u64 flags)
 {
 
@@ -208,6 +239,7 @@ static int create_space_info(struct btrfs_fs_info *info, u64 flags)
 	INIT_LIST_HEAD(&space_info->tickets);
 	INIT_LIST_HEAD(&space_info->priority_tickets);
 	space_info->clamp = 1;
+	btrfs_update_space_info_chunk_size(space_info, calc_chunk_size(info, flags));
 
 	if (btrfs_is_zoned(info))
 		space_info->bg_reclaim_threshold = BTRFS_DEFAULT_ZONED_RECLAIM_THRESH;
fs/btrfs/space-info.h +4 -0
index a803e29bd781..137206b8049f 100644
@@ -23,6 +23,8 @@ struct btrfs_space_info {
 	u64 max_extent_size;	/* This will hold the maximum extent size of
 				   the space info if we had an ENOSPC in the
 				   allocator. */
+	/* Chunk size in bytes */
+	u64 chunk_size;
 
 	/*
 	 * Once a block group drops below this threshold (percents) we'll
@@ -121,6 +123,8 @@ void btrfs_update_space_info(struct btrfs_fs_info *info, u64 flags,
 			     u64 total_bytes, u64 bytes_used,
 			     u64 bytes_readonly, u64 bytes_zone_unusable,
 			     struct btrfs_space_info **space_info);
+void btrfs_update_space_info_chunk_size(struct btrfs_space_info *space_info,
+					u64 chunk_size);
 struct btrfs_space_info *btrfs_find_space_info(struct btrfs_fs_info *info,
 					       u64 flags);
 u64 __pure btrfs_space_info_used(struct btrfs_space_info *s_info,
fs/btrfs/volumes.c +9 -19
index 659575526e9f..4bc97e7d8e46 100644
@@ -5091,26 +5091,16 @@ static void init_alloc_chunk_ctl_policy_regular(
 				struct btrfs_fs_devices *fs_devices,
 				struct alloc_chunk_ctl *ctl)
 {
-	u64 type = ctl->type;
+	struct btrfs_space_info *space_info;
 
-	if (type & BTRFS_BLOCK_GROUP_DATA) {
-		ctl->max_stripe_size = SZ_1G;
-		ctl->max_chunk_size = BTRFS_MAX_DATA_CHUNK_SIZE;
-	} else if (type & BTRFS_BLOCK_GROUP_METADATA) {
-		/* For larger filesystems, use larger metadata chunks */
-		if (fs_devices->total_rw_bytes > 50ULL * SZ_1G)
-			ctl->max_stripe_size = SZ_1G;
-		else
-			ctl->max_stripe_size = SZ_256M;
-		ctl->max_chunk_size = ctl->max_stripe_size;
-	} else if (type & BTRFS_BLOCK_GROUP_SYSTEM) {
-		ctl->max_stripe_size = SZ_32M;
-		ctl->max_chunk_size = 2 * ctl->max_stripe_size;
-		ctl->devs_max = min_t(int, ctl->devs_max,
-				      BTRFS_MAX_DEVS_SYS_CHUNK);
-	} else {
-		BUG();
-	}
+	space_info = btrfs_find_space_info(fs_devices->fs_info, ctl->type);
+	ASSERT(space_info);
+
+	ctl->max_chunk_size = READ_ONCE(space_info->chunk_size);
+	ctl->max_stripe_size = ctl->max_chunk_size;
+
+	if (ctl->type & BTRFS_BLOCK_GROUP_SYSTEM)
+		ctl->devs_max = min_t(int, ctl->devs_max, BTRFS_MAX_DEVS_SYS_CHUNK);
 
 	/* We don't want a chunk larger than 10% of writable space */
 	ctl->max_chunk_size = min(div_factor(fs_devices->total_rw_bytes, 1),
Signature · 1 line
2.35.1
author maintainer
#1029 lore raw

PATCH 1027/1095 btrfs: zoned: introduce space_info->active_total_bytes

In reply to Greg Kroah-Hartman

From: Naohiro Aota <[email protected]>

[ Upstream commit 6a921de589926a350634e6e279f43fa5b9dbf5ba ]

The active_total_bytes, like the total_bytes, accounts for the total bytes of active block groups in the space_info.

With an introduction of active_total_bytes, we can check if the reserved bytes can be written to the block groups without activating a new block group. The check is necessary for metadata allocation on zoned filesystem. We cannot finish a block group, which may require waiting for the current transaction, from the metadata allocation context. Instead, we need to ensure the ongoing allocation (reserved bytes) fits in active block groups.

---
 fs/btrfs/block-group.c | 12 +++++++++---
 fs/btrfs/space-info.c  | 41 ++++++++++++++++++++++++++++++++---------
 fs/btrfs/space-info.h  |  4 +++-
 fs/btrfs/zoned.c       |  6 ++++++
 4 files changed, 50 insertions(+), 13 deletions(-)
fs/btrfs/block-group.c +9 -3
index 1deca5164c23..88f59a2e4113 100644
@@ -1033,8 +1033,13 @@ int btrfs_remove_block_group(struct btrfs_trans_handle *trans,
 			< block_group->zone_unusable);
 		WARN_ON(block_group->space_info->disk_total
 			< block_group->length * factor);
+		WARN_ON(block_group->zone_is_active &&
+			block_group->space_info->active_total_bytes
+			< block_group->length);
 	}
 	block_group->space_info->total_bytes -= block_group->length;
+	if (block_group->zone_is_active)
+		block_group->space_info->active_total_bytes -= block_group->length;
 	block_group->space_info->bytes_readonly -=
 		(block_group->length - block_group->zone_unusable);
 	block_group->space_info->bytes_zone_unusable -=
@@ -2102,7 +2107,8 @@ static int read_one_block_group(struct btrfs_fs_info *info,
 	trace_btrfs_add_block_group(info, cache, 0);
 	btrfs_update_space_info(info, cache->flags, cache->length,
 				cache->used, cache->bytes_super,
-				cache->zone_unusable, &space_info);
+				cache->zone_unusable, cache->zone_is_active,
+				&space_info);
 
 	cache->space_info = space_info;
 
@@ -2172,7 +2178,7 @@ static int fill_dummy_bgs(struct btrfs_fs_info *fs_info)
 		}
 
 		btrfs_update_space_info(fs_info, bg->flags, em->len, em->len,
-					0, 0, &space_info);
+					0, 0, false, &space_info);
 		bg->space_info = space_info;
 		link_block_group(bg);
 
@@ -2553,7 +2559,7 @@ struct btrfs_block_group *btrfs_make_block_group(struct btrfs_trans_handle *tran
 	trace_btrfs_add_block_group(fs_info, cache, 1);
 	btrfs_update_space_info(fs_info, cache->flags, size, bytes_used,
 				cache->bytes_super, cache->zone_unusable,
-				&cache->space_info);
+				cache->zone_is_active, &cache->space_info);
 	btrfs_update_global_block_rsv(fs_info);
 
 	link_block_group(cache);
fs/btrfs/space-info.c +32 -9

103 lines not on this page · raw message

fs/btrfs/space-info.h +3 -1
index 137206b8049f..b8cee27df213 100644
@@ -17,6 +17,8 @@ struct btrfs_space_info {
 	u64 bytes_may_use;	/* number of bytes that may be used for
 				   delalloc/allocations */
 	u64 bytes_readonly;	/* total bytes that are read only */
+	/* Total bytes in the space, but only accounts active block groups. */
+	u64 active_total_bytes;
 	u64 bytes_zone_unusable;	/* total bytes that are unusable until
 					   resetting the device zone */
 
@@ -122,7 +124,7 @@ int btrfs_init_space_info(struct btrfs_fs_info *fs_info);
 void btrfs_update_space_info(struct btrfs_fs_info *info, u64 flags,
 			     u64 total_bytes, u64 bytes_used,
 			     u64 bytes_readonly, u64 bytes_zone_unusable,
-			     struct btrfs_space_info **space_info);
+			     bool active, struct btrfs_space_info **space_info);
 void btrfs_update_space_info_chunk_size(struct btrfs_space_info *space_info,
 					u64 chunk_size);
 struct btrfs_space_info *btrfs_find_space_info(struct btrfs_fs_info *info,
fs/btrfs/zoned.c +6 -0
index 170681797283..2ffc6d50d20d 100644
@@ -1841,6 +1841,7 @@ struct btrfs_device *btrfs_zoned_get_device(struct btrfs_fs_info *fs_info,
 bool btrfs_zone_activate(struct btrfs_block_group *block_group)
 {
 	struct btrfs_fs_info *fs_info = block_group->fs_info;
+	struct btrfs_space_info *space_info = block_group->space_info;
 	struct map_lookup *map;
 	struct btrfs_device *device;
 	u64 physical;
@@ -1852,6 +1853,7 @@ bool btrfs_zone_activate(struct btrfs_block_group *block_group)
 
 	map = block_group->physical_map;
 
+	spin_lock(&space_info->lock);
 	spin_lock(&block_group->lock);
 	if (block_group->zone_is_active) {
 		ret = true;
@@ -1880,7 +1882,10 @@ bool btrfs_zone_activate(struct btrfs_block_group *block_group)
 
 	/* Successfully activated all the zones */
 	block_group->zone_is_active = 1;
+	space_info->active_total_bytes += block_group->length;
 	spin_unlock(&block_group->lock);
+	btrfs_try_granting_tickets(fs_info, space_info);
+	spin_unlock(&space_info->lock);
 
 	/* For the active block group list */
 	btrfs_get_block_group(block_group);
@@ -1893,6 +1898,7 @@ bool btrfs_zone_activate(struct btrfs_block_group *block_group)
 
 out_unlock:
 	spin_unlock(&block_group->lock);
+	spin_unlock(&space_info->lock);
 	return ret;
 }
 
Signature · 1 line
2.35.1
author maintainer
#1030 lore raw

PATCH 1028/1095 btrfs: zoned: activate metadata block group on flush_space

In reply to Greg Kroah-Hartman

From: Naohiro Aota <[email protected]>

[ Upstream commit b0931513913633044ed6e3800334c28433c007b0 ]

For metadata space on zoned filesystem, reaching ALLOC_CHUNK{,_FORCE} means we don't have enough space left in the active_total_bytes. Before allocating a new chunk, we can try to activate an existing block group in this case.

Also, allocating a chunk is not enough to grant a ticket for metadata space on zoned filesystem we need to activate the block group to increase the active_total_bytes.

btrfs_zoned_activate_one_bg() implements the activation feature. It will activate a block group by (maybe) finishing a block group. It will give up activating a block group if it cannot finish any block group.

Fixes afba2bc036b0 ("btrfs: zoned: implement active zone tracking")
---
 fs/btrfs/space-info.c | 30 ++++++++++++++++++++++++
 fs/btrfs/zoned.c      | 53 +++++++++++++++++++++++++++++++++++++++++++
 fs/btrfs/zoned.h      | 10 ++++++++
 3 files changed, 93 insertions(+)
fs/btrfs/space-info.c +30 -0
index 4867199cf983..104cbc901c0e 100644
@@ -9,6 +9,7 @@
 #include "ordered-data.h"
 #include "transaction.h"
 #include "block-group.h"
+#include "zoned.h"
 
 /*
  * HOW DOES SPACE RESERVATION WORK
@@ -724,6 +725,18 @@ static void flush_space(struct btrfs_fs_info *fs_info,
 		break;
 	case ALLOC_CHUNK:
 	case ALLOC_CHUNK_FORCE:
+		/*
+		 * For metadata space on zoned filesystem, reaching here means we
+		 * don't have enough space left in active_total_bytes. Try to
+		 * activate a block group first, because we may have inactive
+		 * block group already allocated.
+		 */
+		ret = btrfs_zoned_activate_one_bg(fs_info, space_info, false);
+		if (ret < 0)
+			break;
+		else if (ret == 1)
+			break;
+
 		trans = btrfs_join_transaction(root);
 		if (IS_ERR(trans)) {
 			ret = PTR_ERR(trans);
@@ -734,6 +747,23 @@ static void flush_space(struct btrfs_fs_info *fs_info,
 				(state == ALLOC_CHUNK) ? CHUNK_ALLOC_NO_FORCE :
 					CHUNK_ALLOC_FORCE);
 		btrfs_end_transaction(trans);
+
+		/*
+		 * For metadata space on zoned filesystem, allocating a new chunk
+		 * is not enough. We still need to activate the block * group.
+		 * Active the newly allocated block group by (maybe) finishing
+		 * a block group.
+		 */
+		if (ret == 1) {
+			ret = btrfs_zoned_activate_one_bg(fs_info, space_info, true);
+			/*
+			 * Revert to the original ret regardless we could finish
+			 * one block group or not.
+			 */
+			if (ret >= 0)
+				ret = 1;
+		}
+
 		if (ret > 0 || ret == -ENOSPC)
 			ret = 0;
 		break;
fs/btrfs/zoned.c +53 -0

56 lines not on this page · raw message

fs/btrfs/zoned.h +10 -0
index 0740458894ac..1cac32266276 100644
@@ -80,6 +80,8 @@ void btrfs_free_zone_cache(struct btrfs_fs_info *fs_info);
 void btrfs_zoned_release_data_reloc_bg(struct btrfs_fs_info *fs_info, u64 logical,
 				       u64 length);
 int btrfs_zone_finish_one_bg(struct btrfs_fs_info *fs_info);
+int btrfs_zoned_activate_one_bg(struct btrfs_fs_info *fs_info,
+				struct btrfs_space_info *space_info, bool do_finish);
 #else /* CONFIG_BLK_DEV_ZONED */
 static inline int btrfs_get_dev_zone(struct btrfs_device *device, u64 pos,
 				     struct blk_zone *zone)
@@ -250,6 +252,14 @@ static inline int btrfs_zone_finish_one_bg(struct btrfs_fs_info *fs_info)
 	return 1;
 }
 
+static inline int btrfs_zoned_activate_one_bg(struct btrfs_fs_info *fs_info,
+					      struct btrfs_space_info *space_info,
+					      bool do_finish)
+{
+	/* Consider all the block groups are active */
+	return 0;
+}
+
 #endif
 
 static inline bool btrfs_dev_is_sequential(struct btrfs_device *device, u64 pos)
Signature · 1 line
2.35.1
author maintainer
#1031 lore raw

PATCH 1029/1095 btrfs: zoned: activate necessary block group

In reply to Greg Kroah-Hartman

From: Naohiro Aota <[email protected]>

[ Upstream commit b6a98021e4019c562a23ad151a7e40adfa9f91e5 ]

There are two places where allocating a chunk is not enough. These two places are trying to ensure the space by allocating a chunk. To meet the condition for active_total_bytes, we also need to activate a block group there.

Fixes afba2bc036b0 ("btrfs: zoned: implement active zone tracking")
---
 fs/btrfs/block-group.c | 16 ++++++++++++++++
 1 file changed, 16 insertions(+)
fs/btrfs/block-group.c +16 -0
index 88f59a2e4113..0c7fe3142d7c 100644
@@ -2659,6 +2659,14 @@ int btrfs_inc_block_group_ro(struct btrfs_block_group *cache,
 	ret = btrfs_chunk_alloc(trans, alloc_flags, CHUNK_ALLOC_FORCE);
 	if (ret < 0)
 		goto out;
+	/*
+	 * We have allocated a new chunk. We also need to activate that chunk to
+	 * grant metadata tickets for zoned filesystem.
+	 */
+	ret = btrfs_zoned_activate_one_bg(fs_info, cache->space_info, true);
+	if (ret < 0)
+		goto out;
+
 	ret = inc_block_group_ro(cache, 0);
 	if (ret == -ETXTBSY)
 		goto unlock_out;
@@ -3853,6 +3861,14 @@ static void reserve_chunk_space(struct btrfs_trans_handle *trans,
 		if (IS_ERR(bg)) {
 			ret = PTR_ERR(bg);
 		} else {
+			/*
+			 * We have a new chunk. We also need to activate it for
+			 * zoned filesystem.
+			 */
+			ret = btrfs_zoned_activate_one_bg(fs_info, info, true);
+			if (ret < 0)
+				return;
+
 			/*
 			 * If we fail to add the chunk item here, we end up
 			 * trying again at phase 2 of chunk allocation, at
Signature · 1 line
2.35.1
author maintainer
#1032 lore raw

PATCH 1030/1095 btrfs: zoned: write out partially allocated region

In reply to Greg Kroah-Hartman

From: Naohiro Aota <[email protected]>

[ Upstream commit 898793d992c23dac6126a6a94ad893eae1a2c9df ]

cow_file_range() works in an all-or-nothing way: if it fails to allocate an extent for a part of the given region, it gives up all the region including the successfully allocated parts. On cow_file_range(), run_delalloc_zoned() writes data for the region only when it successfully allocate all the region.

This all-or-nothing allocation and write-out are problematic when available space in all the block groups are get tight with the active zone restriction. btrfs_reserve_extent() try hard to utilize the left space in the active block groups and gives up finally and fails with -ENOSPC. However, if we send IOs for the successfully allocated region, we can finish a zone and can continue on the rest of the allocation on a newly allocated block group.

This patch implements the partial write-out for run_delalloc_zoned(). With this patch applied, cow_file_range() returns -EAGAIN to tell the caller to do something to progress the further allocation, and tells the successfully allocated region with done_offset. Furthermore, the zoned extent allocator returns -EAGAIN to tell cow_file_range() going back to the caller side.

Actually, we still need to wait for an IO to complete to continue the allocation. The next patch implements that part.

Fixes afba2bc036b0 ("btrfs: zoned: implement active zone tracking")
---
 fs/btrfs/extent-tree.c | 10 +++++++
 fs/btrfs/inode.c       | 63 ++++++++++++++++++++++++++++++++----------
 2 files changed, 59 insertions(+), 14 deletions(-)
fs/btrfs/extent-tree.c +10 -0
index 56185541e188..eee68a6f2be7 100644
@@ -4015,6 +4015,16 @@ static int can_allocate_chunk_zoned(struct btrfs_fs_info *fs_info,
 	if (ffe_ctl->max_extent_size >= ffe_ctl->min_alloc_size)
 		return -ENOSPC;
 
+	/*
+	 * Even min_alloc_size is not left in any block groups. Since we cannot
+	 * activate a new block group, allocating it may not help. Let's tell a
+	 * caller to try again and hope it progress something by writing some
+	 * parts of the region. That is only possible for data block groups,
+	 * where a part of the region can be written.
+	 */
+	if (ffe_ctl->flags & BTRFS_BLOCK_GROUP_DATA)
+		return -EAGAIN;
+
 	/*
 	 * We cannot activate a new block group and no enough space left in any
 	 * block groups. So, allocating a new block group may not help. But,
fs/btrfs/inode.c +49 -14

109 lines not on this page · raw message

Signature · 1 line
2.35.1
author maintainer
#1033 lore raw

PATCH 1031/1095 btrfs: zoned: wait until zone is finished when allocation didnt progress

In reply to Greg Kroah-Hartman

From: Naohiro Aota <[email protected]>

[ Upstream commit 2ce543f478433a0eec0f72090d7e814f1d53d456 ]

When the allocated position doesn't progress, we cannot submit IOs to finish a block group, but there should be ongoing IOs that will finish a block group. So, in that case, we wait for a zone to be finished and retry the allocation after that.

Introduce a new flag BTRFS_FS_NEED_ZONE_FINISH for fs_info->flags to indicate we need a zone finish to have proceeded. The flag is set when the allocator detected it cannot activate a new block group. And, it is cleared once a zone is finished.

Fixes afba2bc036b0 ("btrfs: zoned: implement active zone tracking")
---
 fs/btrfs/ctree.h   | 5 +++++
 fs/btrfs/disk-io.c | 1 +
 fs/btrfs/inode.c   | 9 +++++++--
 fs/btrfs/zoned.c   | 6 ++++++
 4 files changed, 19 insertions(+), 2 deletions(-)
fs/btrfs/ctree.h +5 -0
index 97f5a3d320ff..76fbe4cf2a28 100644
@@ -635,6 +635,9 @@ enum {
 	/* Indicate we have half completed snapshot deletions pending. */
 	BTRFS_FS_UNFINISHED_DROPS,
 
+	/* Indicate we have to finish a zone to do next allocation. */
+	BTRFS_FS_NEED_ZONE_FINISH,
+
 #if BITS_PER_LONG == 32
 	/* Indicate if we have error/warn message printed on 32bit systems */
 	BTRFS_FS_32BIT_ERROR,
@@ -1074,6 +1077,8 @@ struct btrfs_fs_info {
 
 	spinlock_t zone_active_bgs_lock;
 	struct list_head zone_active_bgs;
+	/* Waiters when BTRFS_FS_NEED_ZONE_FINISH is set */
+	wait_queue_head_t zone_finish_wait;
 
 #ifdef CONFIG_BTRFS_FS_REF_VERIFY
 	spinlock_t ref_verify_lock;
fs/btrfs/disk-io.c +1 -0
index bf5c6ac67e87..59fa7bf3a2e5 100644
@@ -3239,6 +3239,7 @@ void btrfs_init_fs_info(struct btrfs_fs_info *fs_info)
 	init_waitqueue_head(&fs_info->transaction_blocked_wait);
 	init_waitqueue_head(&fs_info->async_submit_wait);
 	init_waitqueue_head(&fs_info->delayed_iputs_wait);
+	init_waitqueue_head(&fs_info->zone_finish_wait);
 
 	/* Usable values until the real ones are cached from the superblock */
 	fs_info->nodesize = 4096;
fs/btrfs/inode.c +7 -2
index 9753fc47e488..64d310ecbb84 100644
@@ -1606,8 +1606,13 @@ static noinline int run_delalloc_zoned(struct btrfs_inode *inode,
 		if (ret == 0)
 			done_offset = end;
 
-		if (done_offset == start)
-			return -ENOSPC;
+		if (done_offset == start) {
+			struct btrfs_fs_info *info = inode->root->fs_info;
+
+			wait_var_event(&info->zone_finish_wait,
+				       !test_bit(BTRFS_FS_NEED_ZONE_FINISH, &info->flags));
+			continue;
+		}
 
 		if (!locked_page_done) {
 			__set_page_dirty_nobuffers(locked_page);
fs/btrfs/zoned.c +6 -0
index 0c2d81b0e3d3..45e29b8c705c 100644
@@ -1993,6 +1993,9 @@ int btrfs_zone_finish(struct btrfs_block_group *block_group)
 	/* For active_bg_list */
 	btrfs_put_block_group(block_group);
 
+	clear_bit(BTRFS_FS_NEED_ZONE_FINISH, &fs_info->flags);
+	wake_up_all(&fs_info->zone_finish_wait);
+
 	return 0;
 }
 
@@ -2021,6 +2024,9 @@ bool btrfs_can_activate_zone(struct btrfs_fs_devices *fs_devices, u64 flags)
 	}
 	mutex_unlock(&fs_info->chunk_mutex);
 
+	if (!ret)
+		set_bit(BTRFS_FS_NEED_ZONE_FINISH, &fs_info->flags);
+
 	return ret;
 }
 
Signature · 1 line
2.35.1
author maintainer
#1034 lore raw

PATCH 1032/1095 intel_idle: Add AlderLake support

In reply to Greg Kroah-Hartman

From: Zhang Rui <[email protected]>

[ Upstream commit d1cf8bbfed1edc5108220342ab39e4544d55fbc3 ]

Similar to SPR, the C1 and C1E states on ADL are mutually exclusive. Only one of them can be enabled at a time.

But contrast to SPR, which usually has a strong latency requirement as a Xeon processor, C1E is preferred on ADL for better energy efficiency.

Add custom C-state tables for ADL with both C1 and C1E, and

1. Enable the "C1E promotion" bit in MSR_IA32_POWER_CTL and mark C1 with the CPUIDLE_FLAG_UNUSABLE flag, so C1 is not available by default.

2. Add support for the "preferred_cstates" module parameter, so that users can choose to use C1 instead of C1E by booting with "intel_idle.preferred_cstates=2".

Separate custom C-state tables are introduced for the ADL mobile and desktop processors, because of the exit latency differences between these two variants, especially with respect to PC10.

[ rjw: Changelog edits, code rearrangement ]
---
 drivers/idle/intel_idle.c | 133 ++++++++++++++++++++++++++++++++++++++
 1 file changed, 133 insertions(+)
drivers/idle/intel_idle.c +133 -0
index 47b68c6071be..907700d1e78e 100644
@@ -811,6 +811,106 @@ static struct cpuidle_state icx_cstates[] __initdata = {
 		.enter = NULL }
 };
 
+/*
+ * On AlderLake C1 has to be disabled if C1E is enabled, and vice versa.
+ * C1E is enabled only if "C1E promotion" bit is set in MSR_IA32_POWER_CTL.
+ * But in this case there is effectively no C1, because C1 requests are
+ * promoted to C1E. If the "C1E promotion" bit is cleared, then both C1
+ * and C1E requests end up with C1, so there is effectively no C1E.
+ *
+ * By default we enable C1E and disable C1 by marking it with
+ * 'CPUIDLE_FLAG_UNUSABLE'.
+ */
+static struct cpuidle_state adl_cstates[] __initdata = {
+	{
+		.name = "C1",
+		.desc = "MWAIT 0x00",
+		.flags = MWAIT2flg(0x00) | CPUIDLE_FLAG_UNUSABLE,
+		.exit_latency = 1,
+		.target_residency = 1,
+		.enter = &intel_idle,
+		.enter_s2idle = intel_idle_s2idle, },
+	{
+		.name = "C1E",
+		.desc = "MWAIT 0x01",
+		.flags = MWAIT2flg(0x01) | CPUIDLE_FLAG_ALWAYS_ENABLE,
+		.exit_latency = 2,
+		.target_residency = 4,
+		.enter = &intel_idle,
+		.enter_s2idle = intel_idle_s2idle, },
+	{
+		.name = "C6",
+		.desc = "MWAIT 0x20",
+		.flags = MWAIT2flg(0x20) | CPUIDLE_FLAG_TLB_FLUSHED,
+		.exit_latency = 220,
+		.target_residency = 600,
+		.enter = &intel_idle,
+		.enter_s2idle = intel_idle_s2idle, },
+	{
+		.name = "C8",
+		.desc = "MWAIT 0x40",
+		.flags = MWAIT2flg(0x40) | CPUIDLE_FLAG_TLB_FLUSHED,
+		.exit_latency = 280,
+		.target_residency = 800,
+		.enter = &intel_idle,
+		.enter_s2idle = intel_idle_s2idle, },
+	{
+		.name = "C10",
+		.desc = "MWAIT 0x60",
+		.flags = MWAIT2flg(0x60) | CPUIDLE_FLAG_TLB_FLUSHED,
+		.exit_latency = 680,
+		.target_residency = 2000,
+		.enter = &intel_idle,
+		.enter_s2idle = intel_idle_s2idle, },
+	{
+		.enter = NULL }
+};
+
+static struct cpuidle_state adl_l_cstates[] __initdata = {
+	{
+		.name = "C1",
+		.desc = "MWAIT 0x00",
+		.flags = MWAIT2flg(0x00) | CPUIDLE_FLAG_UNUSABLE,
+		.exit_latency = 1,
+		.target_residency = 1,
+		.enter = &intel_idle,
+		.enter_s2idle = intel_idle_s2idle, },
+	{
+		.name = "C1E",
+		.desc = "MWAIT 0x01",
+		.flags = MWAIT2flg(0x01) | CPUIDLE_FLAG_ALWAYS_ENABLE,
+		.exit_latency = 2,
+		.target_residency = 4,
+		.enter = &intel_idle,
+		.enter_s2idle = intel_idle_s2idle, },
+	{
+		.name = "C6",
+		.desc = "MWAIT 0x20",
+		.flags = MWAIT2flg(0x20) | CPUIDLE_FLAG_TLB_FLUSHED,
+		.exit_latency = 170,
+		.target_residency = 500,
+		.enter = &intel_idle,
+		.enter_s2idle = intel_idle_s2idle, },
+	{
+		.name = "C8",
+		.desc = "MWAIT 0x40",
+		.flags = MWAIT2flg(0x40) | CPUIDLE_FLAG_TLB_FLUSHED,
+		.exit_latency = 200,
+		.target_residency = 600,
+		.enter = &intel_idle,
+		.enter_s2idle = intel_idle_s2idle, },
+	{
+		.name = "C10",
+		.desc = "MWAIT 0x60",
+		.flags = MWAIT2flg(0x60) | CPUIDLE_FLAG_TLB_FLUSHED,
+		.exit_latency = 230,
+		.target_residency = 700,
+		.enter = &intel_idle,
+		.enter_s2idle = intel_idle_s2idle, },
+	{
+		.enter = NULL }
+};
+
 /*
  * On Sapphire Rapids Xeon C1 has to be disabled if C1E is enabled, and vice
  * versa. On SPR C1E is enabled only if "C1E promotion" bit is set in
@@ -1194,6 +1294,14 @@ static const struct idle_cpu idle_cpu_icx __initconst = {
 	.use_acpi = true,
 };
 
+static const struct idle_cpu idle_cpu_adl __initconst = {
+	.state_table = adl_cstates,
+};
+
+static const struct idle_cpu idle_cpu_adl_l __initconst = {
+	.state_table = adl_l_cstates,
+};
+
 static const struct idle_cpu idle_cpu_spr __initconst = {
 	.state_table = spr_cstates,
 	.disable_promotion_to_c1e = true,
@@ -1262,6 +1370,8 @@ static const struct x86_cpu_id intel_idle_ids[] __initconst = {
 	X86_MATCH_INTEL_FAM6_MODEL(SKYLAKE_X,		&idle_cpu_skx),
 	X86_MATCH_INTEL_FAM6_MODEL(ICELAKE_X,		&idle_cpu_icx),
 	X86_MATCH_INTEL_FAM6_MODEL(ICELAKE_D,		&idle_cpu_icx),
+	X86_MATCH_INTEL_FAM6_MODEL(ALDERLAKE,		&idle_cpu_adl),
+	X86_MATCH_INTEL_FAM6_MODEL(ALDERLAKE_L,		&idle_cpu_adl_l),
 	X86_MATCH_INTEL_FAM6_MODEL(SAPPHIRERAPIDS_X,	&idle_cpu_spr),
 	X86_MATCH_INTEL_FAM6_MODEL(XEON_PHI_KNL,	&idle_cpu_knl),
 	X86_MATCH_INTEL_FAM6_MODEL(XEON_PHI_KNM,	&idle_cpu_knl),
@@ -1620,6 +1730,25 @@ static void __init skx_idle_state_table_update(void)
 	}
 }
 
+/**
+ * adl_idle_state_table_update - Adjust AlderLake idle states table.
+ */
+static void __init adl_idle_state_table_update(void)
+{
+	/* Check if user prefers C1 over C1E. */
+	if (preferred_states_mask & BIT(1) && !(preferred_states_mask & BIT(2))) {
+		cpuidle_state_table[0].flags &= ~CPUIDLE_FLAG_UNUSABLE;
+		cpuidle_state_table[1].flags |= CPUIDLE_FLAG_UNUSABLE;
+
+		/* Disable C1E by clearing the "C1E promotion" bit. */
+		c1e_promotion = C1E_PROMOTION_DISABLE;
+		return;
+	}
+
+	/* Make sure C1E is enabled by default */
+	c1e_promotion = C1E_PROMOTION_ENABLE;
+}
+
 /**
  * spr_idle_state_table_update - Adjust Sapphire Rapids idle states table.
  */
@@ -1689,6 +1818,10 @@ static void __init intel_idle_init_cstates_icpu(struct cpuidle_driver *drv)
 	case INTEL_FAM6_SAPPHIRERAPIDS_X:
 		spr_idle_state_table_update();
 		break;
+	case INTEL_FAM6_ALDERLAKE:
+	case INTEL_FAM6_ALDERLAKE_L:
+		adl_idle_state_table_update();
+		break;
 	}
 
 	for (cstate = 0; cstate < CPUIDLE_STATE_MAX; ++cstate) {
Signature · 1 line
2.35.1
author maintainer
#1035 lore raw

PATCH 1033/1095 intel_idle: make SPR C1 and C1E be independent

In reply to Greg Kroah-Hartman

From: Artem Bityutskiy <[email protected]>

[ Upstream commit 1548fac47a114b42063def551eb152a536ed9697 ]

This patch partially reverts the changes made by the following commit:

da0e58c038e6 intel_idle: add 'preferred_cstates' module argument

As that commit describes, on early Sapphire Rapids Xeon platforms the C1 and C1E states were mutually exclusive, so that users could only have either C1 and C6, or C1E and C6.

However, Intel firmware engineers managed to remove this limitation and make C1 and C1E to be completely independent, just like on previous Xeon platforms.

Therefore, this patch:
* Removes commentary describing the old, and now non-existing SPR C1E limitation.
* Marks SPR C1E as available by default.
* Removes the 'preferred_cstates' parameter handling for SPR. Both C1 and C1E will be available regardless of 'preferred_cstates' value.

We expect that all SPR systems are shipping with new firmware, which includes the C1/C1E improvement.

---
 drivers/idle/intel_idle.c | 24 +-----------------------
 1 file changed, 1 insertion(+), 23 deletions(-)
drivers/idle/intel_idle.c +1 -23
index 907700d1e78e..9515a3146dc9 100644
@@ -911,16 +911,6 @@ static struct cpuidle_state adl_l_cstates[] __initdata = {
 		.enter = NULL }
 };
 
-/*
- * On Sapphire Rapids Xeon C1 has to be disabled if C1E is enabled, and vice
- * versa. On SPR C1E is enabled only if "C1E promotion" bit is set in
- * MSR_IA32_POWER_CTL. But in this case there effectively no C1, because C1
- * requests are promoted to C1E. If the "C1E promotion" bit is cleared, then
- * both C1 and C1E requests end up with C1, so there is effectively no C1E.
- *
- * By default we enable C1 and disable C1E by marking it with
- * 'CPUIDLE_FLAG_UNUSABLE'.
- */
 static struct cpuidle_state spr_cstates[] __initdata = {
 	{
 		.name = "C1",
@@ -933,8 +923,7 @@ static struct cpuidle_state spr_cstates[] __initdata = {
 	{
 		.name = "C1E",
 		.desc = "MWAIT 0x01",
-		.flags = MWAIT2flg(0x01) | CPUIDLE_FLAG_ALWAYS_ENABLE |
-					   CPUIDLE_FLAG_UNUSABLE,
+		.flags = MWAIT2flg(0x01) | CPUIDLE_FLAG_ALWAYS_ENABLE,
 		.exit_latency = 2,
 		.target_residency = 4,
 		.enter = &intel_idle,
@@ -1756,17 +1745,6 @@ static void __init spr_idle_state_table_update(void)
 {
 	unsigned long long msr;
 
-	/* Check if user prefers C1E over C1. */
-	if ((preferred_states_mask & BIT(2)) &&
-	    !(preferred_states_mask & BIT(1))) {
-		/* Disable C1 and enable C1E. */
-		spr_cstates[0].flags |= CPUIDLE_FLAG_UNUSABLE;
-		spr_cstates[1].flags &= ~CPUIDLE_FLAG_UNUSABLE;
-
-		/* Enable C1E using the "C1E promotion" bit. */
-		c1e_promotion = C1E_PROMOTION_ENABLE;
-	}
-
 	/*
 	 * By default, the C6 state assumes the worst-case scenario of package
 	 * C6. However, if PC6 is disabled, we update the numbers to match
Signature · 1 line
2.35.1
author maintainer
#1036 lore raw

PATCH 1034/1095 ACPI: CPPC: Do not prevent CPPC from working in the future

In reply to Greg Kroah-Hartman

From: Rafael J. Wysocki <[email protected]>

[ Upstream commit 4f4179fcf420873002035cf1941d844c9e0e7cb3 ]

There is a problem with the current revision checks in is_cppc_supported() that they essentially prevent the CPPC support from working if a new _CPC package format revision being a proper superset of the v3 and only causing _CPC to return a package with more entries (while retaining the types and meaning of the entries defined by the v3) is introduced in the future and used by the platform firmware.

In that case, as long as the number of entries in the _CPC return package is at least CPPC_V3_NUM_ENT, it should be perfectly fine to use the v3 support code and disregard the additional package entries added by the new package format revision.

For this reason, drop is_cppc_supported() altogether, put the revision checks directly into acpi_cppc_processor_probe() so they are easier to follow and rework them to take the case mentioned above into account.

Fixes 4773e77cdc9b ("ACPI / CPPC: Add support for CPPC v3")
Cc 4.18+ <[email protected]> # 4.18+
---
 drivers/acpi/cppc_acpi.c | 54 ++++++++++++++++++----------------------
 include/acpi/cppc_acpi.h |  2 +-
 2 files changed, 25 insertions(+), 31 deletions(-)
drivers/acpi/cppc_acpi.c +24 -30
index b8e26b6b5523..35d894674eba 100644
@@ -600,33 +600,6 @@ static int pcc_data_alloc(int pcc_ss_id)
 	return 0;
 }
 
-/* Check if CPPC revision + num_ent combination is supported */
-static bool is_cppc_supported(int revision, int num_ent)
-{
-	int expected_num_ent;
-
-	switch (revision) {
-	case CPPC_V2_REV:
-		expected_num_ent = CPPC_V2_NUM_ENT;
-		break;
-	case CPPC_V3_REV:
-		expected_num_ent = CPPC_V3_NUM_ENT;
-		break;
-	default:
-		pr_debug("Firmware exports unsupported CPPC revision: %d\n",
-			revision);
-		return false;
-	}
-
-	if (expected_num_ent != num_ent) {
-		pr_debug("Firmware exports %d entries. Expected: %d for CPPC rev:%d\n",
-			num_ent, expected_num_ent, revision);
-		return false;
-	}
-
-	return true;
-}
-
 /*
  * An example CPC table looks like the following.
  *
@@ -715,7 +688,6 @@ int acpi_cppc_processor_probe(struct acpi_processor *pr)
 			 cpc_obj->type, pr->id);
 		goto out_free;
 	}
-	cpc_ptr->num_entries = num_ent;
 
 	/* Second entry should be revision. */
 	cpc_obj = &out_obj->package.elements[1];
@@ -726,10 +698,32 @@ int acpi_cppc_processor_probe(struct acpi_processor *pr)
 			 cpc_obj->type, pr->id);
 		goto out_free;
 	}
-	cpc_ptr->version = cpc_rev;
 
-	if (!is_cppc_supported(cpc_rev, num_ent))
+	if (cpc_rev < CPPC_V2_REV) {
+		pr_debug("Unsupported _CPC Revision (%d) for CPU:%d\n", cpc_rev,
+			 pr->id);
+		goto out_free;
+	}
+
+	/*
+	 * Disregard _CPC if the number of entries in the return pachage is not
+	 * as expected, but support future revisions being proper supersets of
+	 * the v3 and only causing more entries to be returned by _CPC.
+	 */
+	if ((cpc_rev == CPPC_V2_REV && num_ent != CPPC_V2_NUM_ENT) ||
+	    (cpc_rev == CPPC_V3_REV && num_ent != CPPC_V3_NUM_ENT) ||
+	    (cpc_rev > CPPC_V3_REV && num_ent <= CPPC_V3_NUM_ENT)) {
+		pr_debug("Unexpected number of _CPC return package entries (%d) for CPU:%d\n",
+			 num_ent, pr->id);
 		goto out_free;
+	}
+	if (cpc_rev > CPPC_V3_REV) {
+		num_ent = CPPC_V3_NUM_ENT;
+		cpc_rev = CPPC_V3_REV;
+	}
+
+	cpc_ptr->num_entries = num_ent;
+	cpc_ptr->version = cpc_rev;
 
 	/* Iterate through remaining entries in _CPC */
 	for (i = 2; i < num_ent; i++) {
include/acpi/cppc_acpi.h +1 -1
index 181907349b49..a76f8c6b732d 100644
@@ -17,7 +17,7 @@
 #include <acpi/pcc.h>
 #include <acpi/processor.h>
 
-/* Support CPPCv2 and CPPCv3  */
+/* CPPCv2 and CPPCv3 support */
 #define CPPC_V2_REV	2
 #define CPPC_V3_REV	3
 #define CPPC_V2_NUM_ENT	21
Signature · 1 line
2.35.1
author maintainer
#1037 lore raw

PATCH 1035/1095 powerpc/powernv/kvm: Use darn for H_RANDOM on Power9

In reply to Greg Kroah-Hartman

From: Jason A. Donenfeld <[email protected]>

[ Upstream commit 7ef3d06f1bc4a5e62273726f3dc2bd258ae1c71f ]

The existing logic in KVM to support guests calling H_RANDOM only works on Power8, because it looks for an RNG in the device tree, but on Power9 we just use darn.

In addition the existing code needs to work in real mode, so we have the special cased powernv_get_random_real_mode() to deal with that.

Instead just have KVM call ppc_md.get_random_seed(), and do the real mode check inside of there, that way we use whatever RNG is available, including darn on Power9.

Fixes e928e9cb3601 ("KVM: PPC: Book3S HV: Add fast real-mode H_RANDOM implementation.")
[mpe: Rebase on previous commit, update change log appropriately]
---
 arch/powerpc/include/asm/archrandom.h |  5 ----
 arch/powerpc/kvm/book3s_hv_builtin.c  |  7 +++---
 arch/powerpc/platforms/powernv/rng.c  | 36 ++++++---------------------
 3 files changed, 12 insertions(+), 36 deletions(-)
arch/powerpc/include/asm/archrandom.h +0 -5
index 9a53e29680f4..258174304904 100644
@@ -38,12 +38,7 @@ static inline bool __must_check arch_get_random_seed_int(unsigned int *v)
 #endif /* CONFIG_ARCH_RANDOM */
 
 #ifdef CONFIG_PPC_POWERNV
-int powernv_hwrng_present(void);
 int powernv_get_random_long(unsigned long *v);
-int powernv_get_random_real_mode(unsigned long *v);
-#else
-static inline int powernv_hwrng_present(void) { return 0; }
-static inline int powernv_get_random_real_mode(unsigned long *v) { return 0; }
 #endif
 
 #endif /* _ASM_POWERPC_ARCHRANDOM_H */
arch/powerpc/kvm/book3s_hv_builtin.c +4 -3
index 7e52d0beee77..5e4251b76e75 100644
@@ -19,7 +19,7 @@
 #include <asm/interrupt.h>
 #include <asm/kvm_ppc.h>
 #include <asm/kvm_book3s.h>
-#include <asm/archrandom.h>
+#include <asm/machdep.h>
 #include <asm/xics.h>
 #include <asm/xive.h>
 #include <asm/dbell.h>
@@ -176,13 +176,14 @@ EXPORT_SYMBOL_GPL(kvmppc_hcall_impl_hv_realmode);
 
 int kvmppc_hwrng_present(void)
 {
-	return powernv_hwrng_present();
+	return ppc_md.get_random_seed != NULL;
 }
 EXPORT_SYMBOL_GPL(kvmppc_hwrng_present);
 
 long kvmppc_rm_h_random(struct kvm_vcpu *vcpu)
 {
-	if (powernv_get_random_real_mode(&vcpu->arch.regs.gpr[4]))
+	if (ppc_md.get_random_seed &&
+	    ppc_md.get_random_seed(&vcpu->arch.regs.gpr[4]))
 		return H_SUCCESS;
 
 	return H_HARDWARE;
arch/powerpc/platforms/powernv/rng.c +8 -28
index 2287c9cd0cd5..d19305292e1e 100644
@@ -29,15 +29,6 @@ struct powernv_rng {
 
 static DEFINE_PER_CPU(struct powernv_rng *, powernv_rng);
 
-int powernv_hwrng_present(void)
-{
-	struct powernv_rng *rng;
-
-	rng = get_cpu_var(powernv_rng);
-	put_cpu_var(rng);
-	return rng != NULL;
-}
-
 static unsigned long rng_whiten(struct powernv_rng *rng, unsigned long val)
 {
 	unsigned long parity;
@@ -58,19 +49,6 @@ static unsigned long rng_whiten(struct powernv_rng *rng, unsigned long val)
 	return val;
 }
 
-int powernv_get_random_real_mode(unsigned long *v)
-{
-	struct powernv_rng *rng;
-
-	rng = raw_cpu_read(powernv_rng);
-	if (!rng)
-		return 0;
-
-	*v = rng_whiten(rng, __raw_rm_readq(rng->regs_real));
-
-	return 1;
-}
-
 static int powernv_get_random_darn(unsigned long *v)
 {
 	unsigned long val;
@@ -107,12 +85,14 @@ int powernv_get_random_long(unsigned long *v)
 {
 	struct powernv_rng *rng;
 
-	rng = get_cpu_var(powernv_rng);
-
-	*v = rng_whiten(rng, in_be64(rng->regs));
-
-	put_cpu_var(rng);
-
+	if (mfmsr() & MSR_DR) {
+		rng = get_cpu_var(powernv_rng);
+		*v = rng_whiten(rng, in_be64(rng->regs));
+		put_cpu_var(rng);
+	} else {
+		rng = raw_cpu_read(powernv_rng);
+		*v = rng_whiten(rng, __raw_rm_readq(rng->regs_real));
+	}
 	return 1;
 }
 EXPORT_SYMBOL_GPL(powernv_get_random_long);
Signature · 1 line
2.35.1
author maintainer
#1038 lore raw

PATCH 1036/1095 s390/unwind: fix fgraph return address recovery

In reply to Greg Kroah-Hartman

From: Sumanth Korikkar <[email protected]>

[ Upstream commit ded466e1806686794b403ebf031133bbaca76bb2 ]

When HAVE_FUNCTION_GRAPH_RET_ADDR_PTR is defined, the return address to the fgraph caller is recovered by tagging it along with the stack pointer of ftrace stack. This makes the stack unwinding more reliable.

When the fgraph return address is modified to return_to_handler, ftrace_graph_ret_addr tries to restore it to the original value using tagged stack pointer.

Fix this by passing tagged sp to ftrace_graph_ret_addr.

Fixes d81675b60d09 ("s390/unwind: recover kretprobe modified return address in stacktrace")
Cc <[email protected]> # 5.18
---
 arch/s390/include/asm/unwind.h | 2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)
arch/s390/include/asm/unwind.h +1 -1
index 0bf06f1682d8..02462e7100c1 100644
@@ -47,7 +47,7 @@ struct unwind_state {
 static inline unsigned long unwind_recover_ret_addr(struct unwind_state *state,
 						    unsigned long ip)
 {
-	ip = ftrace_graph_ret_addr(state->task, &state->graph_idx, ip, NULL);
+	ip = ftrace_graph_ret_addr(state->task, &state->graph_idx, ip, (void *)state->sp);
 	if (is_kretprobe_trampoline(ip))
 		ip = kretprobe_find_ret_addr(state->task, (void *)state->sp, &state->kr_cur);
 	return ip;
Signature · 1 line
2.35.1
author maintainer
#1039 lore raw

PATCH 1037/1095 KVM: x86/pmu: Introduce the ctrl_mask value for fixed counter

In reply to Greg Kroah-Hartman

From: Like Xu <[email protected]>

[ Upstream commit 2c985527dd8d283e786ad7a67e532ef7f6f00fac ]

The mask value of fixed counter control register should be dynamic adjusted with the number of fixed counters. This patch introduces a variable that includes the reserved bits of fixed counter control registers. This is a generic code refactoring.

Message-Id: <[email protected]>

---
 arch/x86/include/asm/kvm_host.h | 1 +
 arch/x86/kvm/vmx/pmu_intel.c    | 6 +++++-
 2 files changed, 6 insertions(+), 1 deletion(-)
arch/x86/include/asm/kvm_host.h +1 -0
index 57550a427789..35c7a1fce8ea 100644
@@ -508,6 +508,7 @@ struct kvm_pmu {
 	unsigned nr_arch_fixed_counters;
 	unsigned available_event_types;
 	u64 fixed_ctr_ctrl;
+	u64 fixed_ctr_ctrl_mask;
 	u64 global_ctrl;
 	u64 global_status;
 	u64 counter_bitmask[2];
arch/x86/kvm/vmx/pmu_intel.c +5 -1
index b82b6709d7a8..040d598622e3 100644
@@ -395,7 +395,7 @@ static int intel_pmu_set_msr(struct kvm_vcpu *vcpu, struct msr_data *msr_info)
 	case MSR_CORE_PERF_FIXED_CTR_CTRL:
 		if (pmu->fixed_ctr_ctrl == data)
 			return 0;
-		if (!(data & 0xfffffffffffff444ull)) {
+		if (!(data & pmu->fixed_ctr_ctrl_mask)) {
 			reprogram_fixed_counters(pmu, data);
 			return 0;
 		}
@@ -479,6 +479,7 @@ static void intel_pmu_refresh(struct kvm_vcpu *vcpu)
 	struct kvm_cpuid_entry2 *entry;
 	union cpuid10_eax eax;
 	union cpuid10_edx edx;
+	int i;
 
 	pmu->nr_arch_gp_counters = 0;
 	pmu->nr_arch_fixed_counters = 0;
@@ -487,6 +488,7 @@ static void intel_pmu_refresh(struct kvm_vcpu *vcpu)
 	pmu->version = 0;
 	pmu->reserved_bits = 0xffffffff00200000ull;
 	pmu->raw_event_mask = X86_RAW_EVENT_MASK;
+	pmu->fixed_ctr_ctrl_mask = ~0ull;
 
 	entry = kvm_find_cpuid_entry(vcpu, 0xa, 0);
 	if (!entry || !vcpu->kvm->arch.enable_pmu)
@@ -522,6 +524,8 @@ static void intel_pmu_refresh(struct kvm_vcpu *vcpu)
 		setup_fixed_pmc_eventsel(pmu);
 	}
 
+	for (i = 0; i < pmu->nr_arch_fixed_counters; i++)
+		pmu->fixed_ctr_ctrl_mask &= ~(0xbull << (i * 4));
 	pmu->global_ctrl = ((1ull << pmu->nr_arch_gp_counters) - 1) |
 		(((1ull << pmu->nr_arch_fixed_counters) - 1) << INTEL_PMC_IDX_FIXED);
 	pmu->global_ctrl_mask = ~pmu->global_ctrl;
Signature · 1 line
2.35.1
author maintainer
#1040 lore raw

PATCH 1038/1095 KVM: VMX: Mark all PERF_GLOBAL_(OVF)_CTRL bits reserved if theres no vPMU

In reply to Greg Kroah-Hartman

From: Sean Christopherson <[email protected]>

[ Upstream commit 93255bf92939d948bc86d81c6bb70bb0fecc5db1 ]

Mark all MSR_CORE_PERF_GLOBAL_CTRL and MSR_CORE_PERF_GLOBAL_OVF_CTRL bits as reserved if there is no guest vPMU. The nVMX VM-Entry consistency checks do not check for a valid vPMU prior to consuming the masks via kvm_valid_perf_global_ctrl(), i.e. may incorrectly allow a non-zero mask to be loaded via VM-Enter or VM-Exit (well, attempted to be loaded, the actual MSR load will be rejected by intel_is_valid_msr()).

Fixes f5132b01386b ("KVM: Expose a version 2 architectural PMU to a guests")

Message-Id: <[email protected]>

---
 arch/x86/kvm/vmx/pmu_intel.c | 2 ++
 1 file changed, 2 insertions(+)
arch/x86/kvm/vmx/pmu_intel.c +2 -0
index 040d598622e3..cd2d0454f8b0 100644
@@ -488,6 +488,8 @@ static void intel_pmu_refresh(struct kvm_vcpu *vcpu)
 	pmu->version = 0;
 	pmu->reserved_bits = 0xffffffff00200000ull;
 	pmu->raw_event_mask = X86_RAW_EVENT_MASK;
+	pmu->global_ctrl_mask = ~0ull;
+	pmu->global_ovf_ctrl_mask = ~0ull;
 	pmu->fixed_ctr_ctrl_mask = ~0ull;
 
 	entry = kvm_find_cpuid_entry(vcpu, 0xa, 0);
Signature · 1 line
2.35.1
author maintainer
#1041 lore raw

PATCH 1039/1095 KVM: x86/pmu: Ignore pmu->global_ctrl check if vPMU doesnt support global_ctrl

In reply to Greg Kroah-Hartman

From: Like Xu <[email protected]>

[ Upstream commit 98defd2e17803263f49548fea930cfc974d505aa ]

MSR_CORE_PERF_GLOBAL_CTRL is introduced as part of Architecture PMU V2, as indicated by Intel SDM 19.2.2 and the intel_is_valid_msr() function.

So in the absence of global_ctrl support, all PMCs are enabled as AMD does.

Message-Id: <[email protected]>

---
 arch/x86/kvm/vmx/pmu_intel.c | 3 +++
 1 file changed, 3 insertions(+)
arch/x86/kvm/vmx/pmu_intel.c +3 -0
index cd2d0454f8b0..2cbd5f183ab5 100644
@@ -98,6 +98,9 @@ static bool intel_pmc_is_enabled(struct kvm_pmc *pmc)
 {
 	struct kvm_pmu *pmu = pmc_to_pmu(pmc);
 
+	if (pmu->version < 2)
+		return true;
+
 	return test_bit(pmc->idx, (unsigned long *)&pmu->global_ctrl);
 }
 
Signature · 1 line
2.35.1
author maintainer
#1042 lore raw

PATCH 1040/1095 KVM: VMX: Add helper to check if the guest PMU has PERF_GLOBAL_CTRL

In reply to Greg Kroah-Hartman

From: Sean Christopherson <[email protected]>

[ Upstream commit b663f0b5f3d665c261256d1f76e98f077c6e56af ]

Add a helper to check of the guest PMU has PERF_GLOBAL_CTRL, which is unintuitive _and_ diverges from Intel's architecturally defined behavior. Even worse, KVM currently implements the check using two different (but equivalent) checks, _and_ there has been at least one attempt to add a _third_ flavor.

Message-Id: <[email protected]>

---
 arch/x86/kvm/vmx/pmu_intel.c |  4 ++--
 arch/x86/kvm/vmx/vmx.h       | 12 ++++++++++++
 2 files changed, 14 insertions(+), 2 deletions(-)
arch/x86/kvm/vmx/pmu_intel.c +2 -2
index 2cbd5f183ab5..8bd154f8c966 100644
@@ -98,7 +98,7 @@ static bool intel_pmc_is_enabled(struct kvm_pmc *pmc)
 {
 	struct kvm_pmu *pmu = pmc_to_pmu(pmc);
 
-	if (pmu->version < 2)
+	if (!intel_pmu_has_perf_global_ctrl(pmu))
 		return true;
 
 	return test_bit(pmc->idx, (unsigned long *)&pmu->global_ctrl);
@@ -215,7 +215,7 @@ static bool intel_is_valid_msr(struct kvm_vcpu *vcpu, u32 msr)
 	case MSR_CORE_PERF_GLOBAL_STATUS:
 	case MSR_CORE_PERF_GLOBAL_CTRL:
 	case MSR_CORE_PERF_GLOBAL_OVF_CTRL:
-		ret = pmu->version > 1;
+		return intel_pmu_has_perf_global_ctrl(pmu);
 		break;
 	default:
 		ret = get_gp_pmc(pmu, msr, MSR_IA32_PERFCTR0) ||
arch/x86/kvm/vmx/vmx.h +12 -0
index 1e7f9453894b..93aa1f3ea01e 100644
@@ -92,6 +92,18 @@ union vmx_exit_reason {
 	u32 full;
 };
 
+static inline bool intel_pmu_has_perf_global_ctrl(struct kvm_pmu *pmu)
+{
+	/*
+	 * Architecturally, Intel's SDM states that IA32_PERF_GLOBAL_CTRL is
+	 * supported if "CPUID.0AH: EAX[7:0] > 0", i.e. if the PMU version is
+	 * greater than zero.  However, KVM only exposes and emulates the MSR
+	 * to/for the guest if the guest PMU supports at least "Architectural
+	 * Performance Monitoring Version 2".
+	 */
+	return pmu->version > 1;
+}
+
 #define vcpu_to_lbr_desc(vcpu) (&to_vmx(vcpu)->lbr_desc)
 #define vcpu_to_lbr_records(vcpu) (&to_vmx(vcpu)->lbr_desc.records)
 
Signature · 1 line
2.35.1
author maintainer
#1043 lore raw

PATCH 1041/1095 KVM: nVMX: Attempt to load PERF_GLOBAL_CTRL on nVMX xfer iff it exists

In reply to Greg Kroah-Hartman

From: Sean Christopherson <[email protected]>

[ Upstream commit 4496a6f9b45e8cd83343ad86a3984d614e22cf54 ]

Attempt to load PERF_GLOBAL_CTRL during nested VM-Enter/VM-Exit if and only if the MSR exists (according to the guest vCPU model). KVM has very misguided handling of VM_{ENTRY,EXIT}_LOAD_IA32_PERF_GLOBAL_CTRL and attempts to force the nVMX MSR settings to match the vPMU model, i.e. to hide/expose the control based on whether or not the MSR exists from the guest's perspective.

KVM's modifications fail to handle the scenario where the vPMU is hidden from the guest _after_ being exposed to the guest, e.g. by userspace doing multiple KVM_SET_CPUID2 calls, which is allowed if done before any KVM_RUN. nested_vmx_pmu_refresh() is called if and only if there's a recognized vPMU, i.e. KVM will leave the bits in the allow state and then ultimately reject the MSR load and WARN.

KVM should not force the VMX MSRs in the first place. KVM taking control of the MSRs was a misguided attempt at mimicking what commit 5f76f6f5ff96 ("KVM: nVMX: Do not expose MPX VMX controls when guest MPX disabled", 2018-10-01) did for MPX. However, the MPX commit was a workaround for another KVM bug and not something that should be imitated (and it should never been done in the first place).

In other words, KVM's ABI _should_ be that userspace has full control over the MSRs, at which point triggering the WARN that loading the MSR must not fail is trivial.

The intent of the WARN is still valid; KVM has consistency checks to ensure that vmcs12->{guest,host}_ia32_perf_global_ctrl is valid. The problem is that '0' must be considered a valid value at all times, and so the simple/obvious solution is to just not actually load the MSR when it does not exist. It is userspace's responsibility to provide a sane vCPU model, i.e. KVM is well within its ABI and Intel's VMX architecture to skip the loads if the MSR does not exist.

Fixes 03a8871add95 ("KVM: nVMX: Expose load IA32_PERF_GLOBAL_CTRL VM-{Entry,Exit} control")

Message-Id: <[email protected]>

---
 arch/x86/kvm/vmx/nested.c | 4 +++-
 1 file changed, 3 insertions(+), 1 deletion(-)
arch/x86/kvm/vmx/nested.c +3 -1
index aa287302f991..5c62e552082a 100644
@@ -2621,6 +2621,7 @@ static int prepare_vmcs02(struct kvm_vcpu *vcpu, struct vmcs12 *vmcs12,
 		vcpu->arch.walk_mmu->inject_page_fault = vmx_inject_page_fault_nested;
 
 	if ((vmcs12->vm_entry_controls & VM_ENTRY_LOAD_IA32_PERF_GLOBAL_CTRL) &&
+	    intel_pmu_has_perf_global_ctrl(vcpu_to_pmu(vcpu)) &&
 	    WARN_ON_ONCE(kvm_set_msr(vcpu, MSR_CORE_PERF_GLOBAL_CTRL,
 				     vmcs12->guest_ia32_perf_global_ctrl))) {
 		*entry_failure_code = ENTRY_FAIL_DEFAULT;
@@ -4346,7 +4347,8 @@ static void load_vmcs12_host_state(struct kvm_vcpu *vcpu,
 		vmcs_write64(GUEST_IA32_PAT, vmcs12->host_ia32_pat);
 		vcpu->arch.pat = vmcs12->host_ia32_pat;
 	}
-	if (vmcs12->vm_exit_controls & VM_EXIT_LOAD_IA32_PERF_GLOBAL_CTRL)
+	if ((vmcs12->vm_exit_controls & VM_EXIT_LOAD_IA32_PERF_GLOBAL_CTRL) &&
+	    intel_pmu_has_perf_global_ctrl(vcpu_to_pmu(vcpu)))
 		WARN_ON_ONCE(kvm_set_msr(vcpu, MSR_CORE_PERF_GLOBAL_CTRL,
 					 vmcs12->host_ia32_perf_global_ctrl));
 
Signature · 1 line
2.35.1
author maintainer
#1044 lore raw

PATCH 1042/1095 dm raid: fix address sanitizer warning in raid_status

In reply to Greg Kroah-Hartman

From: Mikulas Patocka <[email protected]>

[ Upstream commit 1fbeea217d8f297fe0e0956a1516d14ba97d0396 ]

There is this warning when using a kernel with the address sanitizer and running this testsuite:
https://gitlab.com/cki-project/kernel-tests/-/tree/main/storage/swraid/scsi_raid

==================================================================

BUG: KASAN: slab-out-of-bounds in raid_status+0x1747/0x2820 [dm_raid]
Read of size 4 at addr ffff888079d2c7e8 by task lvcreate/13319
CPU: 0 PID: 13319 Comm: lvcreate Not tainted 5.18.0-0.rc3.<snip> #1
Hardware name: Red Hat KVM, BIOS 0.5.1 01/01/2011
Call Trace:
 <TASK>
 dump_stack_lvl+0x6a/0x9c
 print_address_description.constprop.0+0x1f/0x1e0
 print_report.cold+0x55/0x244
 kasan_report+0xc9/0x100
 raid_status+0x1747/0x2820 [dm_raid]
 dm_ima_measure_on_table_load+0x4b8/0xca0 [dm_mod]
 table_load+0x35c/0x630 [dm_mod]
 ctl_ioctl+0x411/0x630 [dm_mod]
 dm_ctl_ioctl+0xa/0x10 [dm_mod]
 __x64_sys_ioctl+0x12a/0x1a0
 do_syscall_64+0x5b/0x80

The warning is caused by reading conf->max_nr_stripes in raid_status. The code in raid_status reads mddev->private, casts it to struct r5conf and reads the entry max_nr_stripes.

However, if we have different raid type than 4/5/6, mddev->private doesn't point to struct r5conf; it may point to struct r0conf, struct r1conf, struct r10conf or struct mpconf. If we cast a pointer to one of these structs to struct r5conf, we will be reading invalid memory and KASAN warns about it.

Fix this bug by reading struct r5conf only if raid type is 4, 5 or 6.

---
 drivers/md/dm-raid.c | 2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)
drivers/md/dm-raid.c +1 -1
index e362a7471512..92e6b731f9d6 100644
@@ -3514,7 +3514,7 @@ static void raid_status(struct dm_target *ti, status_type_t type,
 {
 	struct raid_set *rs = ti->private;
 	struct mddev *mddev = &rs->md;
-	struct r5conf *conf = mddev->private;
+	struct r5conf *conf = rs_is_raid456(rs) ? mddev->private : NULL;
 	int i, max_nr_stripes = conf ? conf->max_nr_stripes : 0;
 	unsigned long recovery;
 	unsigned int raid_param_cnt = 1; /* at least 1 for chunksize */
Signature · 1 line
2.35.1
author maintainer
#1045 lore raw

PATCH 1043/1095 dm raid: fix address sanitizer warning in raid_resume

In reply to Greg Kroah-Hartman

From: Mikulas Patocka <[email protected]>

[ Upstream commit 7dad24db59d2d2803576f2e3645728866a056dab ]

There is a KASAN warning in raid_resume when running the lvm test lvconvert-raid.sh. The reason for the warning is that mddev->raid_disks is greater than rs->raid_disks, so the loop touches one entry beyond the allocated length.

---
 drivers/md/dm-raid.c | 2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)
drivers/md/dm-raid.c +1 -1
index 92e6b731f9d6..a55d6f6f294b 100644
@@ -3824,7 +3824,7 @@ static void attempt_restore_of_faulty_devices(struct raid_set *rs)
 
 	memset(cleared_failed_devices, 0, sizeof(cleared_failed_devices));
 
-	for (i = 0; i < mddev->raid_disks; i++) {
+	for (i = 0; i < rs->raid_disks; i++) {
 		r = &rs->dev[i].rdev;
 		/* HM FIXME: enhance journal device recovery processing */
 		if (test_bit(Journal, &r->flags))
Signature · 1 line
2.35.1
author maintainer
#1046 lore raw

PATCH 1044/1095 mm/damon/reclaim: fix potential memory leak in damon_reclaim_init()

In reply to Greg Kroah-Hartman

From: Jianglei Nie <[email protected]>

[ Upstream commit 188043c7f4f2bd662f2a55957d684fffa543e600 ]

damon_reclaim_init() allocates a memory chunk for ctx with damon_new_ctx(). When damon_select_ops() fails, ctx is not released, which will lead to a memory leak.

We should release the ctx with damon_destroy_ctx() when damon_select_ops() fails to fix the memory leak.

Fixes 4d69c3457821 ("mm/damon/reclaim: use damon_select_ops() instead of damon_{v,p}a_set_operations()")
---
 mm/damon/reclaim.c | 4 +++-
 1 file changed, 3 insertions(+), 1 deletion(-)
mm/damon/reclaim.c +3 -1
index e34c4d0c4d93..11982685508e 100644
@@ -384,8 +384,10 @@ static int __init damon_reclaim_init(void)
 	if (!ctx)
 		return -ENOMEM;
 
-	if (damon_select_ops(ctx, DAMON_OPS_PADDR))
+	if (damon_select_ops(ctx, DAMON_OPS_PADDR)) {
+		damon_destroy_ctx(ctx);
 		return -EINVAL;
+	}
 
 	ctx->callback.after_aggregation = damon_reclaim_after_aggregation;
 
Signature · 1 line
2.35.1
author maintainer
#1047 lore raw

PATCH 1045/1095 hugetlb_cgroup: fix wrong hugetlb cgroup numa stat

In reply to Greg Kroah-Hartman

From: Miaohe Lin <[email protected]>

[ Upstream commit 2727cfe4072a35ce813e3708f74c135de7da8897 ]

We forget to set cft->private for numa stat file. As a result, numa stat of hstates[0] is always showed for all hstates. Encode the hstates index into cft->private to fix this issue.

Fixes f47761999052 ("hugetlb: add hugetlb.*.numa_stat file")
Cc Kees Cook <[email protected]>
Cc Mike Kravetz <[email protected]>
Cc Mina Almasry <[email protected]>
Cc Shakeel Butt <[email protected]>
---
 mm/hugetlb_cgroup.c | 1 +
 1 file changed, 1 insertion(+)
mm/hugetlb_cgroup.c +1 -0
index f9942841df18..c86691c431fd 100644
@@ -772,6 +772,7 @@ static void __init __hugetlb_cgroup_file_dfl_init(int idx)
 	/* Add the numa stat file */
 	cft = &h->cgroup_files_dfl[6];
 	snprintf(cft->name, MAX_CFTYPE_NAME, "%s.numa_stat", buf);
+	cft->private = MEMFILE_PRIVATE(idx, 0);
 	cft->seq_show = hugetlb_cgroup_read_numa_stat;
 	cft->flags = CFTYPE_NOT_ON_ROOT;
 
Signature · 1 line
2.35.1
author maintainer
#1048 lore raw

PATCH 1046/1095 batman-adv: tracing: Use the new __vstring() helper

In reply to Greg Kroah-Hartman

From: Steven Rostedt (Google) <[email protected]>

[ Upstream commit 9abc291812d784bd4a26c01af4ebdbf9f2dbf0bb ]

Instead of open coding a __dynamic_array() with a fixed length (which defeats the purpose of the dynamic array in the first place). Use the new __vstring() helper that will use a va_list and only write enough of the string into the ring buffer that is needed.

Cc Marek Lindner <[email protected]>
Cc Ingo Molnar <[email protected]>
Cc Andrew Morton <[email protected]>
Cc Simon Wunderlich <[email protected]>
Cc Antonio Quartulli <[email protected]>
Cc "David S. Miller" <[email protected]>
Cc Eric Dumazet <[email protected]>
Cc Jakub Kicinski <[email protected]>
Cc Paolo Abeni <[email protected]>
---
 net/batman-adv/trace.h | 9 ++-------
 1 file changed, 2 insertions(+), 7 deletions(-)
net/batman-adv/trace.h +2 -7
index d673ebdd0426..31c8f922651d 100644
@@ -28,8 +28,6 @@
 
 #endif /* CONFIG_BATMAN_ADV_TRACING */
 
-#define BATADV_MAX_MSG_LEN	256
-
 TRACE_EVENT(batadv_dbg,
 
 	    TP_PROTO(struct batadv_priv *bat_priv,
@@ -40,16 +38,13 @@ TRACE_EVENT(batadv_dbg,
 	    TP_STRUCT__entry(
 		    __string(device, bat_priv->soft_iface->name)
 		    __string(driver, KBUILD_MODNAME)
-		    __dynamic_array(char, msg, BATADV_MAX_MSG_LEN)
+		    __vstring(msg, vaf->fmt, vaf->va)
 	    ),
 
 	    TP_fast_assign(
 		    __assign_str(device, bat_priv->soft_iface->name);
 		    __assign_str(driver, KBUILD_MODNAME);
-		    WARN_ON_ONCE(vsnprintf(__get_dynamic_array(msg),
-					   BATADV_MAX_MSG_LEN,
-					   vaf->fmt,
-					   *vaf->va) >= BATADV_MAX_MSG_LEN);
+		    __assign_vstr(msg, vaf->fmt, vaf->va);
 	    ),
 
 	    TP_printk(
Signature · 1 line
2.35.1
author maintainer
#1049 lore raw

PATCH 1047/1095 tracing: Use a struct alignof to determine trace event field alignment

In reply to Greg Kroah-Hartman

From: Steven Rostedt (Google) <[email protected]>

[ Upstream commit 4c3d2f9388d36eb28640a220a6f908328442d873 ]

alignof() gives an alignment of types as they would be as standalone variables. But alignment in structures might be different, and when building the fields of events, the alignment must be the actual alignment otherwise the field offsets may not match what they actually are.

This caused trace-cmd to crash, as libtraceevent did not check if the field offset was bigger than the event. The write_msr and read_msr events on 32 bit had their fields incorrect, because it had a u64 field between two ints. alignof(u64) would give 8, but the u64 field was at a 4 byte alignment.

Define a macro as:

ALIGN_STRUCTFIELD(type) ((int)(offsetof(struct {char a; type b;}, b)))

which gives the actual alignment of types in a structure.

Cc Ingo Molnar <[email protected]>
Cc Andrew Morton <[email protected]>
Cc Peter Zijlstra <[email protected]>
Cc Thomas Gleixner <[email protected]>
Cc Masami Hiramatsu <[email protected]>
Fixes 04ae87a52074e ("ftrace: Rework event_create_dir()")
---
 include/trace/stages/stage4_event_fields.h | 8 +++++---
 1 file changed, 5 insertions(+), 3 deletions(-)
include/trace/stages/stage4_event_fields.h +5 -3
index c3790ec7a453..80d34f396555 100644
@@ -2,16 +2,18 @@
 
 /* Stage 4 definitions for creating trace events */
 
+#define ALIGN_STRUCTFIELD(type) ((int)(offsetof(struct {char a; type b;}, b)))
+
 #undef __field_ext
 #define __field_ext(_type, _item, _filter_type) {			\
 	.type = #_type, .name = #_item,					\
-	.size = sizeof(_type), .align = __alignof__(_type),		\
+	.size = sizeof(_type), .align = ALIGN_STRUCTFIELD(_type),	\
 	.is_signed = is_signed_type(_type), .filter_type = _filter_type },
 
 #undef __field_struct_ext
 #define __field_struct_ext(_type, _item, _filter_type) {		\
 	.type = #_type, .name = #_item,					\
-	.size = sizeof(_type), .align = __alignof__(_type),		\
+	.size = sizeof(_type), .align = ALIGN_STRUCTFIELD(_type),	\
 	0, .filter_type = _filter_type },
 
 #undef __field
@@ -23,7 +25,7 @@
 #undef __array
 #define __array(_type, _item, _len) {					\
 	.type = #_type"["__stringify(_len)"]", .name = #_item,		\
-	.size = sizeof(_type[_len]), .align = __alignof__(_type),	\
+	.size = sizeof(_type[_len]), .align = ALIGN_STRUCTFIELD(_type),	\
 	.is_signed = is_signed_type(_type), .filter_type = FILTER_OTHER },
 
 #undef __dynamic_array
Signature · 1 line
2.35.1
author maintainer
#1050 lore raw

PATCH 1048/1095 ksmbd: validate length in smb2_write()

In reply to Greg Kroah-Hartman

From: Marios Makassikis <[email protected]>

[ Upstream commit 158a66b245739e15858de42c0ba60fcf3de9b8e6 ]

The SMB2 Write packet contains data that is to be written to a file or to a pipe. Depending on the client, there may be padding between the header and the data field. Currently, the length is validated only in the case padding is present.

Since the DataOffset field always points to the beginning of the data, there is no need to have a special case for padding. By removing this, the length is validated in both cases.

---
 fs/ksmbd/smb2pdu.c |   49 +++++++++++++++++++------------------------------
 1 file changed, 19 insertions(+), 30 deletions(-)
fs/ksmbd/smb2pdu.c +19 -30
@@ -6344,23 +6344,18 @@ static noinline int smb2_write_pipe(stru
 	length = le32_to_cpu(req->Length);
 	id = req->VolatileFileId;
 
-	if (le16_to_cpu(req->DataOffset) ==
-	    offsetof(struct smb2_write_req, Buffer)) {
-		data_buf = (char *)&req->Buffer[0];
-	} else {
-		if ((u64)le16_to_cpu(req->DataOffset) + length >
-		    get_rfc1002_len(work->request_buf)) {
-			pr_err("invalid write data offset %u, smb_len %u\n",
-			       le16_to_cpu(req->DataOffset),
-			       get_rfc1002_len(work->request_buf));
-			err = -EINVAL;
-			goto out;
-		}
-
-		data_buf = (char *)(((char *)&req->hdr.ProtocolId) +
-				le16_to_cpu(req->DataOffset));
+	if ((u64)le16_to_cpu(req->DataOffset) + length >
+	    get_rfc1002_len(work->request_buf)) {
+		pr_err("invalid write data offset %u, smb_len %u\n",
+		       le16_to_cpu(req->DataOffset),
+		       get_rfc1002_len(work->request_buf));
+		err = -EINVAL;
+		goto out;
 	}
 
+	data_buf = (char *)(((char *)&req->hdr.ProtocolId) +
+			   le16_to_cpu(req->DataOffset));
+
 	rpc_resp = ksmbd_rpc_write(work->sess, id, data_buf, length);
 	if (rpc_resp) {
 		if (rpc_resp->flags == KSMBD_RPC_ENOTIMPLEMENTED) {
@@ -6505,22 +6500,16 @@ int smb2_write(struct ksmbd_work *work)
 
 	if (req->Channel != SMB2_CHANNEL_RDMA_V1 &&
 	    req->Channel != SMB2_CHANNEL_RDMA_V1_INVALIDATE) {
-		if (le16_to_cpu(req->DataOffset) ==
-		    offsetof(struct smb2_write_req, Buffer)) {
-			data_buf = (char *)&req->Buffer[0];
-		} else {
-			if ((u64)le16_to_cpu(req->DataOffset) + length >
-			    get_rfc1002_len(work->request_buf)) {
-				pr_err("invalid write data offset %u, smb_len %u\n",
-				       le16_to_cpu(req->DataOffset),
-				       get_rfc1002_len(work->request_buf));
-				err = -EINVAL;
-				goto out;
-			}
-
-			data_buf = (char *)(((char *)&req->hdr.ProtocolId) +
-					le16_to_cpu(req->DataOffset));
+		if ((u64)le16_to_cpu(req->DataOffset) + length >
+		    get_rfc1002_len(work->request_buf)) {
+			pr_err("invalid write data offset %u, smb_len %u\n",
+			       le16_to_cpu(req->DataOffset),
+			       get_rfc1002_len(work->request_buf));
+			err = -EINVAL;
+			goto out;
 		}
+		data_buf = (char *)(((char *)&req->hdr.ProtocolId) +
+				    le16_to_cpu(req->DataOffset));
 
 		ksmbd_debug(SMB, "flags %u\n", le32_to_cpu(req->Flags));
 		if (le32_to_cpu(req->Flags) & SMB2_WRITEFLAG_WRITE_THROUGH)
author maintainer
#1051 lore raw

PATCH 1049/1095 ksmbd: smbd: change prototypes of RDMA read/write related functions

In reply to Greg Kroah-Hartman

From: Hyunchul Lee <[email protected]>

[ Upstream commit 1807abcf8778bcbbf584fe54da9ccbe9029c49bb ]

Change the prototypes of RDMA read/write operations to accept a pointer and length
of buffer descriptors.

---
 fs/ksmbd/connection.c     | 20 ++++++++++----------
 fs/ksmbd/connection.h     | 27 ++++++++++++++++-----------
 fs/ksmbd/smb2pdu.c        | 23 ++++++++---------------
 fs/ksmbd/transport_rdma.c | 30 +++++++++++++++++-------------
 4 files changed, 51 insertions(+), 49 deletions(-)
fs/ksmbd/connection.c +10 -10
index bc6050b67256..e8f476c5f189 100644
@@ -205,31 +205,31 @@ int ksmbd_conn_write(struct ksmbd_work *work)
 	return 0;
 }
 
-int ksmbd_conn_rdma_read(struct ksmbd_conn *conn, void *buf,
-			 unsigned int buflen, u32 remote_key, u64 remote_offset,
-			 u32 remote_len)
+int ksmbd_conn_rdma_read(struct ksmbd_conn *conn,
+			 void *buf, unsigned int buflen,
+			 struct smb2_buffer_desc_v1 *desc,
+			 unsigned int desc_len)
 {
 	int ret = -EINVAL;
 
 	if (conn->transport->ops->rdma_read)
 		ret = conn->transport->ops->rdma_read(conn->transport,
 						      buf, buflen,
-						      remote_key, remote_offset,
-						      remote_len);
+						      desc, desc_len);
 	return ret;
 }
 
-int ksmbd_conn_rdma_write(struct ksmbd_conn *conn, void *buf,
-			  unsigned int buflen, u32 remote_key,
-			  u64 remote_offset, u32 remote_len)
+int ksmbd_conn_rdma_write(struct ksmbd_conn *conn,
+			  void *buf, unsigned int buflen,
+			  struct smb2_buffer_desc_v1 *desc,
+			  unsigned int desc_len)
 {
 	int ret = -EINVAL;
 
 	if (conn->transport->ops->rdma_write)
 		ret = conn->transport->ops->rdma_write(conn->transport,
 						       buf, buflen,
-						       remote_key, remote_offset,
-						       remote_len);
+						       desc, desc_len);
 	return ret;
 }
 
fs/ksmbd/connection.h +16 -11
index 7a59aacb5daa..98c1cbe45ec9 100644
@@ -122,11 +122,14 @@ struct ksmbd_transport_ops {
 	int (*writev)(struct ksmbd_transport *t, struct kvec *iovs, int niov,
 		      int size, bool need_invalidate_rkey,
 		      unsigned int remote_key);
-	int (*rdma_read)(struct ksmbd_transport *t, void *buf, unsigned int len,
-			 u32 remote_key, u64 remote_offset, u32 remote_len);
-	int (*rdma_write)(struct ksmbd_transport *t, void *buf,
-			  unsigned int len, u32 remote_key, u64 remote_offset,
-			  u32 remote_len);
+	int (*rdma_read)(struct ksmbd_transport *t,
+			 void *buf, unsigned int len,
+			 struct smb2_buffer_desc_v1 *desc,
+			 unsigned int desc_len);
+	int (*rdma_write)(struct ksmbd_transport *t,
+			  void *buf, unsigned int len,
+			  struct smb2_buffer_desc_v1 *desc,
+			  unsigned int desc_len);
 };
 
 struct ksmbd_transport {
@@ -148,12 +151,14 @@ struct ksmbd_conn *ksmbd_conn_alloc(void);
 void ksmbd_conn_free(struct ksmbd_conn *conn);
 bool ksmbd_conn_lookup_dialect(struct ksmbd_conn *c);
 int ksmbd_conn_write(struct ksmbd_work *work);
-int ksmbd_conn_rdma_read(struct ksmbd_conn *conn, void *buf,
-			 unsigned int buflen, u32 remote_key, u64 remote_offset,
-			 u32 remote_len);
-int ksmbd_conn_rdma_write(struct ksmbd_conn *conn, void *buf,
-			  unsigned int buflen, u32 remote_key, u64 remote_offset,
-			  u32 remote_len);
+int ksmbd_conn_rdma_read(struct ksmbd_conn *conn,
+			 void *buf, unsigned int buflen,
+			 struct smb2_buffer_desc_v1 *desc,
+			 unsigned int desc_len);
+int ksmbd_conn_rdma_write(struct ksmbd_conn *conn,
+			  void *buf, unsigned int buflen,
+			  struct smb2_buffer_desc_v1 *desc,
+			  unsigned int desc_len);
 void ksmbd_conn_enqueue_request(struct ksmbd_work *work);
 int ksmbd_conn_try_dequeue_request(struct ksmbd_work *work);
 void ksmbd_conn_init_server_callbacks(struct ksmbd_conn_ops *ops);
fs/ksmbd/smb2pdu.c +8 -15
index 5df87fe18905..8f86b8d6765f 100644
@@ -6132,7 +6132,6 @@ static noinline int smb2_read_pipe(struct ksmbd_work *work)
 static int smb2_set_remote_key_for_rdma(struct ksmbd_work *work,
 					struct smb2_buffer_desc_v1 *desc,
 					__le32 Channel,
-					__le16 ChannelInfoOffset,
 					__le16 ChannelInfoLength)
 {
 	unsigned int i, ch_count;
@@ -6158,7 +6157,8 @@ static int smb2_set_remote_key_for_rdma(struct ksmbd_work *work,
 
 	work->need_invalidate_rkey =
 		(Channel == SMB2_CHANNEL_RDMA_V1_INVALIDATE);
-	work->remote_key = le32_to_cpu(desc->token);
+	if (Channel == SMB2_CHANNEL_RDMA_V1_INVALIDATE)
+		work->remote_key = le32_to_cpu(desc->token);
 	return 0;
 }
 
@@ -6166,14 +6166,12 @@ static ssize_t smb2_read_rdma_channel(struct ksmbd_work *work,
 				      struct smb2_read_req *req, void *data_buf,
 				      size_t length)
 {
-	struct smb2_buffer_desc_v1 *desc =
-		(struct smb2_buffer_desc_v1 *)&req->Buffer[0];
 	int err;
 
 	err = ksmbd_conn_rdma_write(work->conn, data_buf, length,
-				    le32_to_cpu(desc->token),
-				    le64_to_cpu(desc->offset),
-				    le32_to_cpu(desc->length));
+				    (struct smb2_buffer_desc_v1 *)
+				    ((char *)req + le16_to_cpu(req->ReadChannelInfoOffset)),
+				    le16_to_cpu(req->ReadChannelInfoLength));
 	if (err)
 		return err;
 
@@ -6217,7 +6215,6 @@ int smb2_read(struct ksmbd_work *work)
 						   (struct smb2_buffer_desc_v1 *)
 						   ((char *)req + ch_offset),
 						   req->Channel,
-						   req->ReadChannelInfoOffset,
 						   req->ReadChannelInfoLength);
 		if (err)
 			goto out;
@@ -6395,21 +6392,18 @@ static ssize_t smb2_write_rdma_channel(struct ksmbd_work *work,
 				       struct ksmbd_file *fp,
 				       loff_t offset, size_t length, bool sync)
 {
-	struct smb2_buffer_desc_v1 *desc;
 	char *data_buf;
 	int ret;
 	ssize_t nbytes;
 
-	desc = (struct smb2_buffer_desc_v1 *)&req->Buffer[0];
-
 	data_buf = kvmalloc(length, GFP_KERNEL | __GFP_ZERO);
 	if (!data_buf)
 		return -ENOMEM;
 
 	ret = ksmbd_conn_rdma_read(work->conn, data_buf, length,
-				   le32_to_cpu(desc->token),
-				   le64_to_cpu(desc->offset),
-				   le32_to_cpu(desc->length));
+				   (struct smb2_buffer_desc_v1 *)
+				   ((char *)req + le16_to_cpu(req->WriteChannelInfoOffset)),
+				   le16_to_cpu(req->WriteChannelInfoLength));
 	if (ret < 0) {
 		kvfree(data_buf);
 		return ret;
@@ -6461,7 +6455,6 @@ int smb2_write(struct ksmbd_work *work)
 						   (struct smb2_buffer_desc_v1 *)
 						   ((char *)req + ch_offset),
 						   req->Channel,
-						   req->WriteChannelInfoOffset,
 						   req->WriteChannelInfoLength);
 		if (err)
 			goto out;
fs/ksmbd/transport_rdma.c +17 -13
index 3f5d13571694..479d279ee146 100644
@@ -1352,14 +1352,18 @@ static void write_done(struct ib_cq *cq, struct ib_wc *wc)
 	read_write_done(cq, wc, DMA_TO_DEVICE);
 }
 
-static int smb_direct_rdma_xmit(struct smb_direct_transport *t, void *buf,
-				int buf_len, u32 remote_key, u64 remote_offset,
-				u32 remote_len, bool is_read)
+static int smb_direct_rdma_xmit(struct smb_direct_transport *t,
+				void *buf, int buf_len,
+				struct smb2_buffer_desc_v1 *desc,
+				unsigned int desc_len,
+				bool is_read)
 {
 	struct smb_direct_rdma_rw_msg *msg;
 	int ret;
 	DECLARE_COMPLETION_ONSTACK(completion);
 	struct ib_send_wr *first_wr = NULL;
+	u32 remote_key = le32_to_cpu(desc[0].token);
+	u64 remote_offset = le64_to_cpu(desc[0].offset);
 
 	ret = wait_for_credits(t, &t->wait_rw_avail_ops, &t->rw_avail_ops);
 	if (ret < 0)
@@ -1424,22 +1428,22 @@ static int smb_direct_rdma_xmit(struct smb_direct_transport *t, void *buf,
 	return ret;
 }
 
-static int smb_direct_rdma_write(struct ksmbd_transport *t, void *buf,
-				 unsigned int buflen, u32 remote_key,
-				 u64 remote_offset, u32 remote_len)
+static int smb_direct_rdma_write(struct ksmbd_transport *t,
+				 void *buf, unsigned int buflen,
+				 struct smb2_buffer_desc_v1 *desc,
+				 unsigned int desc_len)
 {
 	return smb_direct_rdma_xmit(smb_trans_direct_transfort(t), buf, buflen,
-				    remote_key, remote_offset,
-				    remote_len, false);
+				    desc, desc_len, false);
 }
 
-static int smb_direct_rdma_read(struct ksmbd_transport *t, void *buf,
-				unsigned int buflen, u32 remote_key,
-				u64 remote_offset, u32 remote_len)
+static int smb_direct_rdma_read(struct ksmbd_transport *t,
+				void *buf, unsigned int buflen,
+				struct smb2_buffer_desc_v1 *desc,
+				unsigned int desc_len)
 {
 	return smb_direct_rdma_xmit(smb_trans_direct_transfort(t), buf, buflen,
-				    remote_key, remote_offset,
-				    remote_len, true);
+				    desc, desc_len, true);
 }
 
 static void smb_direct_disconnect(struct ksmbd_transport *t)
Signature · 1 line
2.35.1
author maintainer
#1052 lore raw

PATCH 1050/1095 ksmbd: smbd: introduce read/write credits for RDMA read/write

In reply to Greg Kroah-Hartman

From: Hyunchul Lee <[email protected]>

[ Upstream commit ddbdc861e37c168cf2fb8a7b7477f5d18b4daf76 ]

SMB2_READ/SMB2_WRITE request has to be granted the number of rw credits, the pages the request wants to transfer / the maximum pages which can be registered with one MR to read and write a file.
And allocate enough RDMA resources for the maximum number of rw credits allowed by ksmbd.

---
 fs/ksmbd/transport_rdma.c | 120 ++++++++++++++++++++++----------------
 1 file changed, 71 insertions(+), 49 deletions(-)
fs/ksmbd/transport_rdma.c +71 -49
index 479d279ee146..b44a5e584bac 100644
@@ -80,9 +80,7 @@ static int smb_direct_max_fragmented_recv_size = 1024 * 1024;
 /*  The maximum single-message size which can be received */
 static int smb_direct_max_receive_size = 8192;
 
-static int smb_direct_max_read_write_size = 524224;
-
-static int smb_direct_max_outstanding_rw_ops = 8;
+static int smb_direct_max_read_write_size = 8 * 1024 * 1024;
 
 static LIST_HEAD(smb_direct_device_list);
 static DEFINE_RWLOCK(smb_direct_device_lock);
@@ -147,10 +145,12 @@ struct smb_direct_transport {
 	atomic_t		send_credits;
 	spinlock_t		lock_new_recv_credits;
 	int			new_recv_credits;
-	atomic_t		rw_avail_ops;
+	int			max_rw_credits;
+	int			pages_per_rw_credit;
+	atomic_t		rw_credits;
 
 	wait_queue_head_t	wait_send_credits;
-	wait_queue_head_t	wait_rw_avail_ops;
+	wait_queue_head_t	wait_rw_credits;
 
 	mempool_t		*sendmsg_mempool;
 	struct kmem_cache	*sendmsg_cache;
@@ -377,7 +377,7 @@ static struct smb_direct_transport *alloc_transport(struct rdma_cm_id *cm_id)
 	t->reassembly_queue_length = 0;
 	init_waitqueue_head(&t->wait_reassembly_queue);
 	init_waitqueue_head(&t->wait_send_credits);
-	init_waitqueue_head(&t->wait_rw_avail_ops);
+	init_waitqueue_head(&t->wait_rw_credits);
 
 	spin_lock_init(&t->receive_credit_lock);
 	spin_lock_init(&t->recvmsg_queue_lock);
@@ -984,18 +984,19 @@ static int smb_direct_flush_send_list(struct smb_direct_transport *t,
 }
 
 static int wait_for_credits(struct smb_direct_transport *t,
-			    wait_queue_head_t *waitq, atomic_t *credits)
+			    wait_queue_head_t *waitq, atomic_t *total_credits,
+			    int needed)
 {
 	int ret;
 
 	do {
-		if (atomic_dec_return(credits) >= 0)
+		if (atomic_sub_return(needed, total_credits) >= 0)
 			return 0;
 
-		atomic_inc(credits);
+		atomic_add(needed, total_credits);
 		ret = wait_event_interruptible(*waitq,
-					       atomic_read(credits) > 0 ||
-						t->status != SMB_DIRECT_CS_CONNECTED);
+					       atomic_read(total_credits) >= needed ||
+					       t->status != SMB_DIRECT_CS_CONNECTED);
 
 		if (t->status != SMB_DIRECT_CS_CONNECTED)
 			return -ENOTCONN;
@@ -1016,7 +1017,19 @@ static int wait_for_send_credits(struct smb_direct_transport *t,
 			return ret;
 	}
 
-	return wait_for_credits(t, &t->wait_send_credits, &t->send_credits);
+	return wait_for_credits(t, &t->wait_send_credits, &t->send_credits, 1);
+}
+
+static int wait_for_rw_credits(struct smb_direct_transport *t, int credits)
+{
+	return wait_for_credits(t, &t->wait_rw_credits, &t->rw_credits, credits);
+}
+
+static int calc_rw_credits(struct smb_direct_transport *t,
+			   char *buf, unsigned int len)
+{
+	return DIV_ROUND_UP(get_buf_page_count(buf, len),
+			    t->pages_per_rw_credit);
 }
 
 static int smb_direct_create_header(struct smb_direct_transport *t,
@@ -1332,8 +1345,8 @@ static void read_write_done(struct ib_cq *cq, struct ib_wc *wc,
 		smb_direct_disconnect_rdma_connection(t);
 	}
 
-	if (atomic_inc_return(&t->rw_avail_ops) > 0)
-		wake_up(&t->wait_rw_avail_ops);
+	if (atomic_inc_return(&t->rw_credits) > 0)
+		wake_up(&t->wait_rw_credits);
 
 	rdma_rw_ctx_destroy(&msg->rw_ctx, t->qp, t->qp->port,
 			    msg->sg_list, msg->sgt.nents, dir);
@@ -1364,8 +1377,10 @@ static int smb_direct_rdma_xmit(struct smb_direct_transport *t,
 	struct ib_send_wr *first_wr = NULL;
 	u32 remote_key = le32_to_cpu(desc[0].token);
 	u64 remote_offset = le64_to_cpu(desc[0].offset);
+	int credits_needed;
 
-	ret = wait_for_credits(t, &t->wait_rw_avail_ops, &t->rw_avail_ops);
+	credits_needed = calc_rw_credits(t, buf, buf_len);
+	ret = wait_for_rw_credits(t, credits_needed);
 	if (ret < 0)
 		return ret;
 
@@ -1373,7 +1388,7 @@ static int smb_direct_rdma_xmit(struct smb_direct_transport *t,
 	msg = kmalloc(offsetof(struct smb_direct_rdma_rw_msg, sg_list) +
 		      sizeof(struct scatterlist) * SG_CHUNK_SIZE, GFP_KERNEL);
 	if (!msg) {
-		atomic_inc(&t->rw_avail_ops);
+		atomic_add(credits_needed, &t->rw_credits);
 		return -ENOMEM;
 	}
 
@@ -1382,7 +1397,7 @@ static int smb_direct_rdma_xmit(struct smb_direct_transport *t,
 				     get_buf_page_count(buf, buf_len),
 				     msg->sg_list, SG_CHUNK_SIZE);
 	if (ret) {
-		atomic_inc(&t->rw_avail_ops);
+		atomic_add(credits_needed, &t->rw_credits);
 		kfree(msg);
 		return -ENOMEM;
 	}
@@ -1418,7 +1433,7 @@ static int smb_direct_rdma_xmit(struct smb_direct_transport *t,
 	return 0;
 
 err:
-	atomic_inc(&t->rw_avail_ops);
+	atomic_add(credits_needed, &t->rw_credits);
 	if (first_wr)
 		rdma_rw_ctx_destroy(&msg->rw_ctx, t->qp, t->qp->port,
 				    msg->sg_list, msg->sgt.nents,
@@ -1643,11 +1658,19 @@ static int smb_direct_prepare_negotiation(struct smb_direct_transport *t)
 	return ret;
 }
 
+static unsigned int smb_direct_get_max_fr_pages(struct smb_direct_transport *t)
+{
+	return min_t(unsigned int,
+		     t->cm_id->device->attrs.max_fast_reg_page_list_len,
+		     256);
+}
+
 static int smb_direct_init_params(struct smb_direct_transport *t,
 				  struct ib_qp_cap *cap)
 {
 	struct ib_device *device = t->cm_id->device;
-	int max_send_sges, max_pages, max_rw_wrs, max_send_wrs;
+	int max_send_sges, max_rw_wrs, max_send_wrs;
+	unsigned int max_sge_per_wr, wrs_per_credit;
 
 	/* need 2 more sge. because a SMB_DIRECT header will be mapped,
 	 * and maybe a send buffer could be not page aligned.
@@ -1659,25 +1682,31 @@ static int smb_direct_init_params(struct smb_direct_transport *t,
 		return -EINVAL;
 	}
 
-	/*
-	 * allow smb_direct_max_outstanding_rw_ops of in-flight RDMA
-	 * read/writes. HCA guarantees at least max_send_sge of sges for
-	 * a RDMA read/write work request, and if memory registration is used,
-	 * we need reg_mr, local_inv wrs for each read/write.
+	/* Calculate the number of work requests for RDMA R/W.
+	 * The maximum number of pages which can be registered
+	 * with one Memory region can be transferred with one
+	 * R/W credit. And at least 4 work requests for each credit
+	 * are needed for MR registration, RDMA R/W, local & remote
+	 * MR invalidation.
 	 */
 	t->max_rdma_rw_size = smb_direct_max_read_write_size;
-	max_pages = DIV_ROUND_UP(t->max_rdma_rw_size, PAGE_SIZE) + 1;
-	max_rw_wrs = DIV_ROUND_UP(max_pages, SMB_DIRECT_MAX_SEND_SGES);
-	max_rw_wrs += rdma_rw_mr_factor(device, t->cm_id->port_num,
-			max_pages) * 2;
-	max_rw_wrs *= smb_direct_max_outstanding_rw_ops;
+	t->pages_per_rw_credit = smb_direct_get_max_fr_pages(t);
+	t->max_rw_credits = DIV_ROUND_UP(t->max_rdma_rw_size,
+					 (t->pages_per_rw_credit - 1) *
+					 PAGE_SIZE);
+
+	max_sge_per_wr = min_t(unsigned int, device->attrs.max_send_sge,
+			       device->attrs.max_sge_rd);
+	wrs_per_credit = max_t(unsigned int, 4,
+			       DIV_ROUND_UP(t->pages_per_rw_credit,
+					    max_sge_per_wr) + 1);
+	max_rw_wrs = t->max_rw_credits * wrs_per_credit;
 
 	max_send_wrs = smb_direct_send_credit_target + max_rw_wrs;
 	if (max_send_wrs > device->attrs.max_cqe ||
 	    max_send_wrs > device->attrs.max_qp_wr) {
-		pr_err("consider lowering send_credit_target = %d, or max_outstanding_rw_ops = %d\n",
-		       smb_direct_send_credit_target,
-		       smb_direct_max_outstanding_rw_ops);
+		pr_err("consider lowering send_credit_target = %d\n",
+		       smb_direct_send_credit_target);
 		pr_err("Possible CQE overrun, device reporting max_cqe %d max_qp_wr %d\n",
 		       device->attrs.max_cqe, device->attrs.max_qp_wr);
 		return -EINVAL;
@@ -1712,7 +1741,7 @@ static int smb_direct_init_params(struct smb_direct_transport *t,
 
 	t->send_credit_target = smb_direct_send_credit_target;
 	atomic_set(&t->send_credits, 0);
-	atomic_set(&t->rw_avail_ops, smb_direct_max_outstanding_rw_ops);
+	atomic_set(&t->rw_credits, t->max_rw_credits);
 
 	t->max_send_size = smb_direct_max_send_size;
 	t->max_recv_size = smb_direct_max_receive_size;
@@ -1720,12 +1749,10 @@ static int smb_direct_init_params(struct smb_direct_transport *t,
 
 	cap->max_send_wr = max_send_wrs;
 	cap->max_recv_wr = t->recv_credit_max;
-	cap->max_send_sge = SMB_DIRECT_MAX_SEND_SGES;
+	cap->max_send_sge = max_sge_per_wr;
 	cap->max_recv_sge = SMB_DIRECT_MAX_RECV_SGES;
 	cap->max_inline_data = 0;
-	cap->max_rdma_ctxs =
-		rdma_rw_mr_factor(device, t->cm_id->port_num, max_pages) *
-		smb_direct_max_outstanding_rw_ops;
+	cap->max_rdma_ctxs = t->max_rw_credits;
 	return 0;
 }
 
@@ -1818,7 +1845,8 @@ static int smb_direct_create_qpair(struct smb_direct_transport *t,
 	}
 
 	t->send_cq = ib_alloc_cq(t->cm_id->device, t,
-				 t->send_credit_target, 0, IB_POLL_WORKQUEUE);
+				 smb_direct_send_credit_target + cap->max_rdma_ctxs,
+				 0, IB_POLL_WORKQUEUE);
 	if (IS_ERR(t->send_cq)) {
 		pr_err("Can't create RDMA send CQ\n");
 		ret = PTR_ERR(t->send_cq);
@@ -1827,8 +1855,7 @@ static int smb_direct_create_qpair(struct smb_direct_transport *t,
 	}
 
 	t->recv_cq = ib_alloc_cq(t->cm_id->device, t,
-				 cap->max_send_wr + cap->max_rdma_ctxs,
-				 0, IB_POLL_WORKQUEUE);
+				 t->recv_credit_max, 0, IB_POLL_WORKQUEUE);
 	if (IS_ERR(t->recv_cq)) {
 		pr_err("Can't create RDMA recv CQ\n");
 		ret = PTR_ERR(t->recv_cq);
@@ -1857,17 +1884,12 @@ static int smb_direct_create_qpair(struct smb_direct_transport *t,
 
 	pages_per_rw = DIV_ROUND_UP(t->max_rdma_rw_size, PAGE_SIZE) + 1;
 	if (pages_per_rw > t->cm_id->device->attrs.max_sgl_rd) {
-		int pages_per_mr, mr_count;
-
-		pages_per_mr = min_t(int, pages_per_rw,
-				     t->cm_id->device->attrs.max_fast_reg_page_list_len);
-		mr_count = DIV_ROUND_UP(pages_per_rw, pages_per_mr) *
-			atomic_read(&t->rw_avail_ops);
-		ret = ib_mr_pool_init(t->qp, &t->qp->rdma_mrs, mr_count,
-				      IB_MR_TYPE_MEM_REG, pages_per_mr, 0);
+		ret = ib_mr_pool_init(t->qp, &t->qp->rdma_mrs,
+				      t->max_rw_credits, IB_MR_TYPE_MEM_REG,
+				      t->pages_per_rw_credit, 0);
 		if (ret) {
 			pr_err("failed to init mr pool count %d pages %d\n",
-			       mr_count, pages_per_mr);
+			       t->max_rw_credits, t->pages_per_rw_credit);
 			goto err;
 		}
 	}
Signature · 1 line
2.35.1
author maintainer
#1053 lore raw

PATCH 1051/1095 ksmbd: add smbd max io size parameter

In reply to Greg Kroah-Hartman

From: Namjae Jeon <[email protected]>

[ Upstream commit 65bb45b97b578c8eed1ffa80caec84708df49729 ]

Add 'smbd max io size' parameter to adjust smbd-direct max read/write size.

---
 fs/ksmbd/ksmbd_netlink.h  | 3 ++-
 fs/ksmbd/transport_ipc.c  | 3 +++
 fs/ksmbd/transport_rdma.c | 8 +++++++-
 fs/ksmbd/transport_rdma.h | 6 ++++++
 4 files changed, 18 insertions(+), 2 deletions(-)
fs/ksmbd/ksmbd_netlink.h +2 -1
index ebe6ca08467a..52aa0adeb951 100644
@@ -104,7 +104,8 @@ struct ksmbd_startup_request {
 					 */
 	__u32	sub_auth[3];		/* Subauth value for Security ID */
 	__u32	smb2_max_credits;	/* MAX credits */
-	__u32	reserved[128];		/* Reserved room */
+	__u32	smbd_max_io_size;	/* smbd read write size */
+	__u32	reserved[127];		/* Reserved room */
 	__u32	ifc_list_sz;		/* interfaces list size */
 	__s8	____payload[];
 };
fs/ksmbd/transport_ipc.c +3 -0
index 3ad6881e0f7e..7cb0eeb07c80 100644
@@ -26,6 +26,7 @@
 #include "mgmt/ksmbd_ida.h"
 #include "connection.h"
 #include "transport_tcp.h"
+#include "transport_rdma.h"
 
 #define IPC_WAIT_TIMEOUT	(2 * HZ)
 
@@ -303,6 +304,8 @@ static int ipc_server_config_on_startup(struct ksmbd_startup_request *req)
 		init_smb2_max_trans_size(req->smb2_max_trans);
 	if (req->smb2_max_credits)
 		init_smb2_max_credits(req->smb2_max_credits);
+	if (req->smbd_max_io_size)
+		init_smbd_max_io_size(req->smbd_max_io_size);
 
 	ret = ksmbd_set_netbios_name(req->netbios_name);
 	ret |= ksmbd_set_server_string(req->server_string);
fs/ksmbd/transport_rdma.c +7 -1
index b44a5e584bac..afc66b9765e7 100644
@@ -80,7 +80,7 @@ static int smb_direct_max_fragmented_recv_size = 1024 * 1024;
 /*  The maximum single-message size which can be received */
 static int smb_direct_max_receive_size = 8192;
 
-static int smb_direct_max_read_write_size = 8 * 1024 * 1024;
+static int smb_direct_max_read_write_size = SMBD_DEFAULT_IOSIZE;
 
 static LIST_HEAD(smb_direct_device_list);
 static DEFINE_RWLOCK(smb_direct_device_lock);
@@ -214,6 +214,12 @@ struct smb_direct_rdma_rw_msg {
 	struct scatterlist	sg_list[];
 };
 
+void init_smbd_max_io_size(unsigned int sz)
+{
+	sz = clamp_val(sz, SMBD_MIN_IOSIZE, SMBD_MAX_IOSIZE);
+	smb_direct_max_read_write_size = sz;
+}
+
 static inline int get_buf_page_count(void *buf, int size)
 {
 	return DIV_ROUND_UP((uintptr_t)buf + size, PAGE_SIZE) -
fs/ksmbd/transport_rdma.h +6 -0
index 5567d93a6f96..e7b4e6790fab 100644
@@ -7,6 +7,10 @@
 #ifndef __KSMBD_TRANSPORT_RDMA_H__
 #define __KSMBD_TRANSPORT_RDMA_H__
 
+#define SMBD_DEFAULT_IOSIZE (8 * 1024 * 1024)
+#define SMBD_MIN_IOSIZE (512 * 1024)
+#define SMBD_MAX_IOSIZE (16 * 1024 * 1024)
+
 /* SMB DIRECT negotiation request packet [MS-SMBD] 2.2.1 */
 struct smb_direct_negotiate_req {
 	__le16 min_version;
@@ -52,10 +56,12 @@ struct smb_direct_data_transfer {
 int ksmbd_rdma_init(void);
 void ksmbd_rdma_destroy(void);
 bool ksmbd_rdma_capable_netdev(struct net_device *netdev);
+void init_smbd_max_io_size(unsigned int sz);
 #else
 static inline int ksmbd_rdma_init(void) { return 0; }
 static inline int ksmbd_rdma_destroy(void) { return 0; }
 static inline bool ksmbd_rdma_capable_netdev(struct net_device *netdev) { return false; }
+static inline void init_smbd_max_io_size(unsigned int sz) { }
 #endif
 
 #endif /* __KSMBD_TRANSPORT_RDMA_H__ */
Signature · 1 line
2.35.1
author maintainer
#1054 lore raw

PATCH 1052/1095 ksmbd: fix wrong smbd max read/write size check

In reply to Greg Kroah-Hartman

From: Namjae Jeon <[email protected]>

[ Upstream commit 7a84399e1ce3f5f2fbec3e7dd93459ba25badc2f ]

smb-direct max read/write size can be different with smb2 max read/write size. So smb2_read() can return error by wrong max read/write size check. This patch use smb_direct_max_read_write_size for this check in smb-direct read/write().

---
 fs/ksmbd/smb2pdu.c        | 39 +++++++++++++++++++++++++--------------
 fs/ksmbd/transport_rdma.c |  5 +++++
 fs/ksmbd/transport_rdma.h |  2 ++
 3 files changed, 32 insertions(+), 14 deletions(-)
fs/ksmbd/smb2pdu.c +25 -14
index 8f86b8d6765f..6c8dd718b5db 100644
@@ -6194,6 +6194,8 @@ int smb2_read(struct ksmbd_work *work)
 	size_t length, mincount;
 	ssize_t nbytes = 0, remain_bytes = 0;
 	int err = 0;
+	bool is_rdma_channel = false;
+	unsigned int max_read_size = conn->vals->max_read_size;
 
 	WORK_BUFFERS(work, req, rsp);
 
@@ -6205,6 +6207,11 @@ int smb2_read(struct ksmbd_work *work)
 
 	if (req->Channel == SMB2_CHANNEL_RDMA_V1_INVALIDATE ||
 	    req->Channel == SMB2_CHANNEL_RDMA_V1) {
+		is_rdma_channel = true;
+		max_read_size = get_smbd_max_read_write_size();
+	}
+
+	if (is_rdma_channel == true) {
 		unsigned int ch_offset = le16_to_cpu(req->ReadChannelInfoOffset);
 
 		if (ch_offset < offsetof(struct smb2_read_req, Buffer)) {
@@ -6236,9 +6243,9 @@ int smb2_read(struct ksmbd_work *work)
 	length = le32_to_cpu(req->Length);
 	mincount = le32_to_cpu(req->MinimumCount);
 
-	if (length > conn->vals->max_read_size) {
+	if (length > max_read_size) {
 		ksmbd_debug(SMB, "limiting read size to max size(%u)\n",
-			    conn->vals->max_read_size);
+			    max_read_size);
 		err = -EINVAL;
 		goto out;
 	}
@@ -6270,8 +6277,7 @@ int smb2_read(struct ksmbd_work *work)
 	ksmbd_debug(SMB, "nbytes %zu, offset %lld mincount %zu\n",
 		    nbytes, offset, mincount);
 
-	if (req->Channel == SMB2_CHANNEL_RDMA_V1_INVALIDATE ||
-	    req->Channel == SMB2_CHANNEL_RDMA_V1) {
+	if (is_rdma_channel == true) {
 		/* write data to the client using rdma channel */
 		remain_bytes = smb2_read_rdma_channel(work, req,
 						      work->aux_payload_buf,
@@ -6432,8 +6438,9 @@ int smb2_write(struct ksmbd_work *work)
 	size_t length;
 	ssize_t nbytes;
 	char *data_buf;
-	bool writethrough = false;
+	bool writethrough = false, is_rdma_channel = false;
 	int err = 0;
+	unsigned int max_write_size = work->conn->vals->max_write_size;
 
 	WORK_BUFFERS(work, req, rsp);
 
@@ -6442,8 +6449,17 @@ int smb2_write(struct ksmbd_work *work)
 		return smb2_write_pipe(work);
 	}
 
+	offset = le64_to_cpu(req->Offset);
+	length = le32_to_cpu(req->Length);
+
 	if (req->Channel == SMB2_CHANNEL_RDMA_V1 ||
 	    req->Channel == SMB2_CHANNEL_RDMA_V1_INVALIDATE) {
+		is_rdma_channel = true;
+		max_write_size = get_smbd_max_read_write_size();
+		length = le32_to_cpu(req->RemainingBytes);
+	}
+
+	if (is_rdma_channel == true) {
 		unsigned int ch_offset = le16_to_cpu(req->WriteChannelInfoOffset);
 
 		if (req->Length != 0 || req->DataOffset != 0 ||
@@ -6478,12 +6494,9 @@ int smb2_write(struct ksmbd_work *work)
 		goto out;
 	}
 
-	offset = le64_to_cpu(req->Offset);
-	length = le32_to_cpu(req->Length);
-
-	if (length > work->conn->vals->max_write_size) {
+	if (length > max_write_size) {
 		ksmbd_debug(SMB, "limiting write size to max size(%u)\n",
-			    work->conn->vals->max_write_size);
+			    max_write_size);
 		err = -EINVAL;
 		goto out;
 	}
@@ -6491,8 +6504,7 @@ int smb2_write(struct ksmbd_work *work)
 	if (le32_to_cpu(req->Flags) & SMB2_WRITEFLAG_WRITE_THROUGH)
 		writethrough = true;
 
-	if (req->Channel != SMB2_CHANNEL_RDMA_V1 &&
-	    req->Channel != SMB2_CHANNEL_RDMA_V1_INVALIDATE) {
+	if (is_rdma_channel == false) {
 		if ((u64)le16_to_cpu(req->DataOffset) + length >
 		    get_rfc1002_len(work->request_buf)) {
 			pr_err("invalid write data offset %u, smb_len %u\n",
@@ -6518,8 +6530,7 @@ int smb2_write(struct ksmbd_work *work)
 		/* read data from the client using rdma channel, and
 		 * write the data.
 		 */
-		nbytes = smb2_write_rdma_channel(work, req, fp, offset,
-						 le32_to_cpu(req->RemainingBytes),
+		nbytes = smb2_write_rdma_channel(work, req, fp, offset, length,
 						 writethrough);
 		if (nbytes < 0) {
 			err = (int)nbytes;
fs/ksmbd/transport_rdma.c +5 -0
index afc66b9765e7..c6af8d89b7f7 100644
@@ -220,6 +220,11 @@ void init_smbd_max_io_size(unsigned int sz)
 	smb_direct_max_read_write_size = sz;
 }
 
+unsigned int get_smbd_max_read_write_size(void)
+{
+	return smb_direct_max_read_write_size;
+}
+
 static inline int get_buf_page_count(void *buf, int size)
 {
 	return DIV_ROUND_UP((uintptr_t)buf + size, PAGE_SIZE) -
fs/ksmbd/transport_rdma.h +2 -0
index e7b4e6790fab..77aee4e5c9dc 100644
@@ -57,11 +57,13 @@ int ksmbd_rdma_init(void);
 void ksmbd_rdma_destroy(void);
 bool ksmbd_rdma_capable_netdev(struct net_device *netdev);
 void init_smbd_max_io_size(unsigned int sz);
+unsigned int get_smbd_max_read_write_size(void);
 #else
 static inline int ksmbd_rdma_init(void) { return 0; }
 static inline int ksmbd_rdma_destroy(void) { return 0; }
 static inline bool ksmbd_rdma_capable_netdev(struct net_device *netdev) { return false; }
 static inline void init_smbd_max_io_size(unsigned int sz) { }
+static inline unsigned int get_smbd_max_read_write_size(void) { return 0; }
 #endif
 
 #endif /* __KSMBD_TRANSPORT_RDMA_H__ */
Signature · 1 line
2.35.1
author maintainer
#1055 lore raw

PATCH 1053/1095 ksmbd: prevent out of bound read for SMB2_WRITE

In reply to Greg Kroah-Hartman

From: Hyunchul Lee <[email protected]>

[ Upstream commit ac60778b87e45576d7bfdbd6f53df902654e6f09 ]

OOB read memory can be written to a file,
if DataOffset is 0 and Length is too large
in SMB2_WRITE request of compound request.

To prevent this, when checking the length of the data area of SMB2_WRITE in smb2_get_data_area_len(), let the minimum of DataOffset be the size of SMB2 header + the size of SMB2_WRITE header.

This bug can lead an oops looking something like:

[  798.008715] BUG: KASAN: slab-out-of-bounds in copy_page_from_iter_atomic+0xd3d/0x14b0
[  798.008724] Read of size 252 at addr ffff88800f863e90 by task kworker/0:2/2859
...
[  798.008754] Call Trace:
[  798.008756]  <TASK>
[  798.008759]  dump_stack_lvl+0x49/0x5f
[  798.008764]  print_report.cold+0x5e/0x5cf
[  798.008768]  ? __filemap_get_folio+0x285/0x6d0
[  798.008774]  ? copy_page_from_iter_atomic+0xd3d/0x14b0
[  798.008777]  kasan_report+0xaa/0x120
[  798.008781]  ? copy_page_from_iter_atomic+0xd3d/0x14b0
[  798.008784]  kasan_check_range+0x100/0x1e0
[  798.008788]  memcpy+0x24/0x60
[  798.008792]  copy_page_from_iter_atomic+0xd3d/0x14b0
[  798.008795]  ? pagecache_get_page+0x53/0x160
[  798.008799]  ? iov_iter_get_pages_alloc+0x1590/0x1590
[  798.008803]  ? ext4_write_begin+0xfc0/0xfc0
[  798.008807]  ? current_time+0x72/0x210
[  798.008811]  generic_perform_write+0x2c8/0x530
[  798.008816]  ? filemap_fdatawrite_wbc+0x180/0x180
[  798.008820]  ? down_write+0xb4/0x120
[  798.008824]  ? down_write_killable+0x130/0x130
[  798.008829]  ext4_buffered_write_iter+0x137/0x2c0
[  798.008833]  ext4_file_write_iter+0x40b/0x1490
[  798.008837]  ? __fsnotify_parent+0x275/0xb20
[  798.008842]  ? __fsnotify_update_child_dentry_flags+0x2c0/0x2c0
[  798.008846]  ? ext4_buffered_write_iter+0x2c0/0x2c0
[  798.008851]  __kernel_write+0x3a1/0xa70
[  798.008855]  ? __x64_sys_preadv2+0x160/0x160
[  798.008860]  ? security_file_permission+0x4a/0xa0
[  798.008865]  kernel_write+0xbb/0x360
[  798.008869]  ksmbd_vfs_write+0x27e/0xb90 [ksmbd]
[  798.008881]  ? ksmbd_vfs_read+0x830/0x830 [ksmbd]
[  798.008892]  ? _raw_read_unlock+0x2a/0x50
[  798.008896]  smb2_write+0xb45/0x14e0 [ksmbd]
[  798.008909]  ? __kasan_check_write+0x14/0x20
[  798.008912]  ? _raw_spin_lock_bh+0xd0/0xe0
[  798.008916]  ? smb2_read+0x15e0/0x15e0 [ksmbd]
[  798.008927]  ? memcpy+0x4e/0x60
[  798.008931]  ? _raw_spin_unlock+0x19/0x30
[  798.008934]  ? ksmbd_smb2_check_message+0x16af/0x2350 [ksmbd]
[  798.008946]  ? _raw_spin_lock_bh+0xe0/0xe0
[  798.008950]  handle_ksmbd_work+0x30e/0x1020 [ksmbd]
[  798.008962]  process_one_work+0x778/0x11c0
[  798.008966]  ? _raw_spin_lock_irq+0x8e/0xe0
[  798.008970]  worker_thread+0x544/0x1180
[  798.008973]  ? __cpuidle_text_end+0x4/0x4
[  798.008977]  kthread+0x282/0x320
[  798.008982]  ? process_one_work+0x11c0/0x11c0
[  798.008985]  ? kthread_complete_and_exit+0x30/0x30
[  798.008989]  ret_from_fork+0x1f/0x30
[  798.008995]  </TASK>
Fixes e2f34481b24d ("cifsd: add server-side procedures for SMB3")
---
 fs/ksmbd/smb2misc.c | 7 +++++--
 fs/ksmbd/smb2pdu.c  | 8 +++-----
 2 files changed, 8 insertions(+), 7 deletions(-)
fs/ksmbd/smb2misc.c +5 -2
index 03bcd7ce0c75..6e25ace36568 100644
@@ -131,8 +131,11 @@ static int smb2_get_data_area_len(unsigned int *off, unsigned int *len,
 		*len = le16_to_cpu(((struct smb2_read_req *)hdr)->ReadChannelInfoLength);
 		break;
 	case SMB2_WRITE:
-		if (((struct smb2_write_req *)hdr)->DataOffset) {
-			*off = le16_to_cpu(((struct smb2_write_req *)hdr)->DataOffset);
+		if (((struct smb2_write_req *)hdr)->DataOffset ||
+		    ((struct smb2_write_req *)hdr)->Length) {
+			*off = max_t(unsigned int,
+				     le16_to_cpu(((struct smb2_write_req *)hdr)->DataOffset),
+				     offsetof(struct smb2_write_req, Buffer));
 			*len = le32_to_cpu(((struct smb2_write_req *)hdr)->Length);
 			break;
 		}
fs/ksmbd/smb2pdu.c +3 -5
index 6c8dd718b5db..85a9ed7156ea 100644
@@ -6505,14 +6505,12 @@ int smb2_write(struct ksmbd_work *work)
 		writethrough = true;
 
 	if (is_rdma_channel == false) {
-		if ((u64)le16_to_cpu(req->DataOffset) + length >
-		    get_rfc1002_len(work->request_buf)) {
-			pr_err("invalid write data offset %u, smb_len %u\n",
-			       le16_to_cpu(req->DataOffset),
-			       get_rfc1002_len(work->request_buf));
+		if (le16_to_cpu(req->DataOffset) <
+		    offsetof(struct smb2_write_req, Buffer)) {
 			err = -EINVAL;
 			goto out;
 		}
+
 		data_buf = (char *)(((char *)&req->hdr.ProtocolId) +
 				    le16_to_cpu(req->DataOffset));
 
Signature · 1 line
2.35.1
author maintainer
#1056 lore raw

PATCH 1054/1095 ext4: update s_overhead_clusters in the superblock during an on-line resize

In reply to Greg Kroah-Hartman

From: Theodore Ts'o <[email protected]>

[ Upstream commit de394a86658ffe4e89e5328fd4993abfe41b7435 ]

When doing an online resize, the on-disk superblock on-disk wasn't updated. This means that when the file system is unmounted and remounted, and the on-disk overhead value is non-zero, this would result in the results of statfs(2) to be incorrect.

This was partially fixed by Commits 10b01ee92df5 ("ext4: fix overhead calculation to account for the reserved gdt blocks"), 85d825dbf489 ("ext4: force overhead calculation if the s_overhead_cluster makes no sense"), and eb7054212eac ("ext4: update the cached overhead value in the superblock").

However, since it was too expensive to forcibly recalculate the overhead for bigalloc file systems at every mount, this didn't fix the problem for bigalloc file systems. This commit should address the problem when resizing file systems with the bigalloc feature enabled.

---
 fs/ext4/resize.c | 1 +
 1 file changed, 1 insertion(+)
fs/ext4/resize.c +1 -0
index 8b70a4701293..e5c2713aa11a 100644
@@ -1484,6 +1484,7 @@ static void ext4_update_super(struct super_block *sb,
 	 * Update the fs overhead information
 	 */
 	ext4_calculate_overhead(sb);
+	es->s_overhead_clusters = cpu_to_le32(sbi->s_overhead);
 
 	if (test_opt(sb, DEBUG))
 		printk(KERN_DEBUG "EXT4-fs: added group %u:"
Signature · 1 line
2.35.1
author maintainer
#1057 lore raw

PATCH 1055/1095 ext4: fix extent status tree race in writeback error recovery path

In reply to Greg Kroah-Hartman

From: Eric Whitney <[email protected]>

[ Upstream commit 7f0d8e1d607c1a4fa9a27362a108921d82230874 ]

A race can occur in the unlikely event ext4 is unable to allocate a physical cluster for a delayed allocation in a bigalloc file system during writeback. Failure to allocate a cluster forces error recovery that includes a call to mpage_release_unused_pages(). That function removes any corresponding delayed allocated blocks from the extent status tree. If a new delayed write is in progress on the same cluster simultaneously, resulting in the addition of an new extent containing one or more blocks in that cluster to the extent status tree, delayed block accounting can be thrown off if that delayed write then encounters a similar cluster allocation failure during future writeback.

Write lock the i_data_sem in mpage_release_unused_pages() to fix this problem. Ext4's block/cluster accounting code for bigalloc relies on i_data_sem for mutual exclusion, as is found in the delayed write path, and the locking in mpage_release_unused_pages() is missing.

---
 fs/ext4/inode.c | 7 +++++++
 1 file changed, 7 insertions(+)
fs/ext4/inode.c +7 -0
index beed9e32571c..826e2deb10f8 100644
@@ -1559,7 +1559,14 @@ static void mpage_release_unused_pages(struct mpage_da_data *mpd,
 		ext4_lblk_t start, last;
 		start = index << (PAGE_SHIFT - inode->i_blkbits);
 		last = end << (PAGE_SHIFT - inode->i_blkbits);
+
+		/*
+		 * avoid racing with extent status tree scans made by
+		 * ext4_insert_delayed_block()
+		 */
+		down_write(&EXT4_I(inode)->i_data_sem);
 		ext4_es_remove_extent(inode, start, last - start + 1);
+		up_write(&EXT4_I(inode)->i_data_sem);
 	}
 
 	pagevec_init(&pvec);
Signature · 1 line
2.35.1
author maintainer
#1058 lore raw

PATCH 1056/1095 ext4: add EXT4_INODE_HAS_XATTR_SPACE macro in xattr.h

In reply to Greg Kroah-Hartman

From: Baokun Li <[email protected]>

[ Upstream commit 179b14152dcb6a24c3415200603aebca70ff13af ]

When adding an xattr to an inode, we must ensure that the inode_size is not less than EXT4_GOOD_OLD_INODE_SIZE + extra_isize + pad. Otherwise, the end position may be greater than the start position, resulting in UAF.

---
 fs/ext4/xattr.h | 13 +++++++++++++
 1 file changed, 13 insertions(+)
fs/ext4/xattr.h +13 -0
index 77efb9a627ad..f885f362add4 100644
@@ -95,6 +95,19 @@ struct ext4_xattr_entry {
 
 #define EXT4_ZERO_XATTR_VALUE ((void *)-1)
 
+/*
+ * If we want to add an xattr to the inode, we should make sure that
+ * i_extra_isize is not 0 and that the inode size is not less than
+ * EXT4_GOOD_OLD_INODE_SIZE + extra_isize + pad.
+ *   EXT4_GOOD_OLD_INODE_SIZE   extra_isize header   entry   pad  data
+ * |--------------------------|------------|------|---------|---|-------|
+ */
+#define EXT4_INODE_HAS_XATTR_SPACE(inode)				\
+	((EXT4_I(inode)->i_extra_isize != 0) &&				\
+	 (EXT4_GOOD_OLD_INODE_SIZE + EXT4_I(inode)->i_extra_isize +	\
+	  sizeof(struct ext4_xattr_ibody_header) + EXT4_XATTR_PAD <=	\
+	  EXT4_INODE_SIZE((inode)->i_sb)))
+
 struct ext4_xattr_info {
 	const char *name;
 	const void *value;
Signature · 1 line
2.35.1
author maintainer
#1059 lore raw

PATCH 1057/1095 ext4: fix use-after-free in ext4_xattr_set_entry

In reply to Greg Kroah-Hartman

From: Baokun Li <[email protected]>

[ Upstream commit 67d7d8ad99beccd9fe92d585b87f1760dc9018e3 ]

Hulk Robot reported a issue:
==================================================================

BUG: KASAN: use-after-free in ext4_xattr_set_entry+0x18ab/0x3500
Write of size 4105 at addr ffff8881675ef5f4 by task syz-executor.0/7092

CPU: 1 PID: 7092 Comm: syz-executor.0 Not tainted 4.19.90-dirty #17
Call Trace:
[...]
 memcpy+0x34/0x50 mm/kasan/kasan.c:303
 ext4_xattr_set_entry+0x18ab/0x3500 fs/ext4/xattr.c:1747
 ext4_xattr_ibody_inline_set+0x86/0x2a0 fs/ext4/xattr.c:2205
 ext4_xattr_set_handle+0x940/0x1300 fs/ext4/xattr.c:2386
 ext4_xattr_set+0x1da/0x300 fs/ext4/xattr.c:2498
 __vfs_setxattr+0x112/0x170 fs/xattr.c:149
 __vfs_setxattr_noperm+0x11b/0x2a0 fs/xattr.c:180
 __vfs_setxattr_locked+0x17b/0x250 fs/xattr.c:238
 vfs_setxattr+0xed/0x270 fs/xattr.c:255
 setxattr+0x235/0x330 fs/xattr.c:520
 path_setxattr+0x176/0x190 fs/xattr.c:539
 __do_sys_lsetxattr fs/xattr.c:561 [inline]
 __se_sys_lsetxattr fs/xattr.c:557 [inline]
 __x64_sys_lsetxattr+0xc2/0x160 fs/xattr.c:557
 do_syscall_64+0xdf/0x530 arch/x86/entry/common.c:298
 entry_SYSCALL_64_after_hwframe+0x44/0xa9
RIP: 0033:0x459fe9
RSP: 002b:00007fa5e54b4c08 EFLAGS: 00000246 ORIG_RAX: 00000000000000bd
RAX: ffffffffffffffda RBX: 000000000051bf60 RCX: 0000000000459fe9
RDX: 00000000200003c0 RSI: 0000000020000180 RDI: 0000000020000140
RBP: 000000000051bf60 R08: 0000000000000001 R09: 0000000000000000
R10: 0000000000001009 R11: 0000000000000246 R12: 0000000000000000
R13: 00007ffc73c93fc0 R14: 000000000051bf60 R15: 00007fa5e54b4d80

[...]
==================================================================

Above issue may happen as follows:
-------------------------------------
ext4_xattr_set
ext4_xattr_set_handle
ext4_xattr_ibody_find
>> s->end < s->base
>> no EXT4_STATE_XATTR
>> xattr_check_inode is not executed
ext4_xattr_ibody_set
ext4_xattr_set_entry
>> size_t min_offs = s->end - s->base
>> UAF in memcpy

we can easily reproduce this problem with the following commands:
mkfs.ext4 -F /dev/sda
mount -o debug_want_extra_isize=128 /dev/sda /mnt
touch /mnt/file
setfattr -n user.cat -v seq -s z 4096|tr -d '[:digit:]' /mnt/file

In ext4_xattr_ibody_find, we have the following assignment logic:
header = IHDR(inode, raw_inode)
= raw_inode + EXT4_GOOD_OLD_INODE_SIZE + i_extra_isize
is->s.base = IFIRST(header)
= header + sizeof(struct ext4_xattr_ibody_header)
is->s.end = raw_inode + s_inode_size

In ext4_xattr_set_entry
min_offs = s->end - s->base
= s_inode_size - EXT4_GOOD_OLD_INODE_SIZE - i_extra_isize -
sizeof(struct ext4_xattr_ibody_header)
last = s->first
free = min_offs - ((void *)last - s->base) - sizeof(__u32)
= s_inode_size - EXT4_GOOD_OLD_INODE_SIZE - i_extra_isize -
sizeof(struct ext4_xattr_ibody_header) - sizeof(__u32)

In the calculation formula, all values except s_inode_size and i_extra_size are fixed values. When i_extra_size is the maximum value s_inode_size - EXT4_GOOD_OLD_INODE_SIZE, min_offs is -4 and free is -8. The value overflows. As a result, the preceding issue is triggered when memcpy is executed.

Therefore, when finding xattr or setting xattr, check whether there is space for storing xattr in the inode to resolve this issue.

---
 fs/ext4/xattr.c | 6 ++++--
 1 file changed, 4 insertions(+), 2 deletions(-)
fs/ext4/xattr.c +4 -2
index 042325349098..c3c3194f3ee1 100644
@@ -2176,8 +2176,9 @@ int ext4_xattr_ibody_find(struct inode *inode, struct ext4_xattr_info *i,
 	struct ext4_inode *raw_inode;
 	int error;
 
-	if (EXT4_I(inode)->i_extra_isize == 0)
+	if (!EXT4_INODE_HAS_XATTR_SPACE(inode))
 		return 0;
+
 	raw_inode = ext4_raw_inode(&is->iloc);
 	header = IHDR(inode, raw_inode);
 	is->s.base = is->s.first = IFIRST(header);
@@ -2205,8 +2206,9 @@ int ext4_xattr_ibody_set(handle_t *handle, struct inode *inode,
 	struct ext4_xattr_search *s = &is->s;
 	int error;
 
-	if (EXT4_I(inode)->i_extra_isize == 0)
+	if (!EXT4_INODE_HAS_XATTR_SPACE(inode))
 		return -ENOSPC;
+
 	error = ext4_xattr_set_entry(i, s, handle, inode, false /* is_block */);
 	if (error)
 		return error;
Signature · 1 line
2.35.1
author maintainer
#1060 lore raw

PATCH 1058/1095 ext4: correct max_inline_xattr_value_size computing

In reply to Greg Kroah-Hartman

From: Baokun Li <[email protected]>

[ Upstream commit c9fd167d57133c5b748d16913c4eabc55e531c73 ]

If the ext4 inode does not have xattr space, 0 is returned in the get_max_inline_xattr_value_size function. Otherwise, the function returns a negative value when the inode does not contain EXT4_STATE_XATTR.

---
 fs/ext4/inline.c | 3 +++
 1 file changed, 3 insertions(+)
fs/ext4/inline.c +3 -0
index e9ef5cf30969..84fcd06a8e8a 100644
@@ -35,6 +35,9 @@ static int get_max_inline_xattr_value_size(struct inode *inode,
 	struct ext4_inode *raw_inode;
 	int free, min_offs;
 
+	if (!EXT4_INODE_HAS_XATTR_SPACE(inode))
+		return 0;
+
 	min_offs = EXT4_SB(inode->i_sb)->s_inode_size -
 			EXT4_GOOD_OLD_INODE_SIZE -
 			EXT4_I(inode)->i_extra_isize -
Signature · 1 line
2.35.1
author maintainer
#1061 lore raw

PATCH 1059/1095 ext4: correct the misjudgment in ext4_iget_extra_inode

In reply to Greg Kroah-Hartman

From: Baokun Li <[email protected]>

[ Upstream commit fd7e672ea98b95b9d4c9dae316639f03c16a749d ]

Use the EXT4_INODE_HAS_XATTR_SPACE macro to more accurately determine whether the inode have xattr space.

---
 fs/ext4/inode.c | 3 +--
 1 file changed, 1 insertion(+), 2 deletions(-)
fs/ext4/inode.c +1 -2
index 826e2deb10f8..e478cac3b8f2 100644
@@ -4681,8 +4681,7 @@ static inline int ext4_iget_extra_inode(struct inode *inode,
 	__le32 *magic = (void *)raw_inode +
 			EXT4_GOOD_OLD_INODE_SIZE + ei->i_extra_isize;
 
-	if (EXT4_GOOD_OLD_INODE_SIZE + ei->i_extra_isize + sizeof(__le32) <=
-	    EXT4_INODE_SIZE(inode->i_sb) &&
+	if (EXT4_INODE_HAS_XATTR_SPACE(inode)  &&
 	    *magic == cpu_to_le32(EXT4_XATTR_MAGIC)) {
 		ext4_set_inode_state(inode, EXT4_STATE_XATTR);
 		return ext4_find_inline_data_nolock(inode);
Signature · 1 line
2.35.1
author maintainer
#1062 lore raw

PATCH 1060/1095 ext4: fix warning in ext4_iomap_begin as race between bmap and write

In reply to Greg Kroah-Hartman

From: Ye Bin <[email protected]>

[ Upstream commit 51ae846cff568c8c29921b1b28eb2dfbcd4ac12d ]

We got issue as follows:

------------[ cut here ]------------
WARNING: CPU: 3 PID: 9310 at fs/ext4/inode.c:3441 ext4_iomap_begin+0x182/0x5d0
RIP: 0010:ext4_iomap_begin+0x182/0x5d0
RSP: 0018:ffff88812460fa08 EFLAGS: 00010293
RAX: ffff88811f168000 RBX: 0000000000000000 RCX: ffffffff97793c12
RDX: 0000000000000000 RSI: 0000000000000000 RDI: 0000000000000003
RBP: ffff88812c669160 R08: ffff88811f168000 R09: ffffed10258cd20f
R10: ffff88812c669077 R11: ffffed10258cd20e R12: 0000000000000001
R13: 00000000000000a4 R14: 000000000000000c R15: ffff88812c6691ee
FS:  00007fd0d6ff3740(0000) GS:ffff8883af180000(0000) knlGS:0000000000000000
CS:  0010 DS: 0000 ES: 0000 CR0: 0000000080050033
CR2: 00007fd0d6dda290 CR3: 0000000104a62000 CR4: 00000000000006e0
DR0: 0000000000000000 DR1: 0000000000000000 DR2: 0000000000000000
DR3: 0000000000000000 DR6: 00000000fffe0ff0 DR7: 0000000000000400
Call Trace:
 iomap_apply+0x119/0x570
 iomap_bmap+0x124/0x150
 ext4_bmap+0x14f/0x250
 bmap+0x55/0x80
 do_vfs_ioctl+0x952/0xbd0
 __x64_sys_ioctl+0xc6/0x170
 do_syscall_64+0x33/0x40
 entry_SYSCALL_64_after_hwframe+0x44/0xa9

Above issue may happen as follows:
bmap write
bmap
ext4_bmap

    iomap_bmap
      ext4_iomap_begin
                            ext4_file_write_iter
			      ext4_buffered_write_iter
			        generic_perform_write
				  ext4_da_write_begin
				    ext4_da_write_inline_data_begin
				      ext4_prepare_inline_data
				        ext4_create_inline_data
					  ext4_set_inode_flag(inode,
						EXT4_INODE_INLINE_DATA);
      if (WARN_ON_ONCE(ext4_has_inline_data(inode))) ->trigger bug_on

To solved above issue hold inode lock in ext4_bamp.

---
 fs/ext4/inode.c | 12 +++++++++---
 1 file changed, 9 insertions(+), 3 deletions(-)
fs/ext4/inode.c +9 -3
index e478cac3b8f2..9ef6f41a5250 100644
@@ -3137,13 +3137,15 @@ static sector_t ext4_bmap(struct address_space *mapping, sector_t block)
 {
 	struct inode *inode = mapping->host;
 	journal_t *journal;
+	sector_t ret = 0;
 	int err;
 
+	inode_lock_shared(inode);
 	/*
 	 * We can get here for an inline file via the FIBMAP ioctl
 	 */
 	if (ext4_has_inline_data(inode))
-		return 0;
+		goto out;
 
 	if (mapping_tagged(mapping, PAGECACHE_TAG_DIRTY) &&
 			test_opt(inode->i_sb, DELALLOC)) {
@@ -3182,10 +3184,14 @@ static sector_t ext4_bmap(struct address_space *mapping, sector_t block)
 		jbd2_journal_unlock_updates(journal);
 
 		if (err)
-			return 0;
+			goto out;
 	}
 
-	return iomap_bmap(mapping, block, &ext4_iomap_ops);
+	ret = iomap_bmap(mapping, block, &ext4_iomap_ops);
+
+out:
+	inode_unlock_shared(inode);
+	return ret;
 }
 
 static int ext4_readpage(struct file *file, struct page *page)
Signature · 1 line
2.35.1
author maintainer
#1063 lore raw

PATCH 1061/1095 ext4: check if directory block is within i_size

In reply to Greg Kroah-Hartman

From: Lukas Czerner <[email protected]>

[ Upstream commit 65f8ea4cd57dbd46ea13b41dc8bac03176b04233 ]

Currently ext4 directory handling code implicitly assumes that the directory blocks are always within the i_size. In fact ext4_append() will attempt to allocate next directory block based solely on i_size and the i_size is then appropriately increased after a successful allocation.

However, for this to work it requires i_size to be correct. If, for any reason, the directory inode i_size is corrupted in a way that the directory tree refers to a valid directory block past i_size, we could end up corrupting parts of the directory tree structure by overwriting already used directory blocks when modifying the directory.

Fix it by catching the corruption early in __ext4_read_dirblock().

Addresses Red-Hat-Bugzilla: #2070205
CVE: CVE-2022-1184

---
 fs/ext4/namei.c | 7 +++++++
 1 file changed, 7 insertions(+)
fs/ext4/namei.c +7 -0
index 4f0420b1ff3e..2bc3e4b27204 100644
@@ -110,6 +110,13 @@ static struct buffer_head *__ext4_read_dirblock(struct inode *inode,
 	struct ext4_dir_entry *dirent;
 	int is_dx_block = 0;
 
+	if (block >= inode->i_size) {
+		ext4_error_inode(inode, func, line, block,
+		       "Attempting to read directory block (%u) that is past i_size (%llu)",
+		       block, inode->i_size);
+		return ERR_PTR(-EFSCORRUPTED);
+	}
+
 	if (ext4_simulate_fail(inode->i_sb, EXT4_SIM_DIRBLOCK_EIO))
 		bh = ERR_PTR(-EIO);
 	else
Signature · 1 line
2.35.1
author maintainer
#1064 lore raw

PATCH 1062/1095 ext4: make sure ext4_append() always allocates new block

In reply to Greg Kroah-Hartman

From: Lukas Czerner <[email protected]>

[ Upstream commit b8a04fe77ef1360fbf73c80fddbdfeaa9407ed1b ]

ext4_append() must always allocate a new block, otherwise we run the risk of overwriting existing directory block corrupting the directory tree in the process resulting in all manner of problems later on.

Add a sanity check to see if the logical block is already allocated and error out if it is.

---
 fs/ext4/namei.c | 16 ++++++++++++++++
 1 file changed, 16 insertions(+)
fs/ext4/namei.c +16 -0
index 2bc3e4b27204..13b6265848c2 100644
@@ -54,6 +54,7 @@ static struct buffer_head *ext4_append(handle_t *handle,
 					struct inode *inode,
 					ext4_lblk_t *block)
 {
+	struct ext4_map_blocks map;
 	struct buffer_head *bh;
 	int err;
 
@@ -63,6 +64,21 @@ static struct buffer_head *ext4_append(handle_t *handle,
 		return ERR_PTR(-ENOSPC);
 
 	*block = inode->i_size >> inode->i_sb->s_blocksize_bits;
+	map.m_lblk = *block;
+	map.m_len = 1;
+
+	/*
+	 * We're appending new directory block. Make sure the block is not
+	 * allocated yet, otherwise we will end up corrupting the
+	 * directory.
+	 */
+	err = ext4_map_blocks(NULL, inode, &map, 0);
+	if (err < 0)
+		return ERR_PTR(err);
+	if (err) {
+		EXT4_ERROR_INODE(inode, "Logical block already allocated");
+		return ERR_PTR(-EFSCORRUPTED);
+	}
 
 	bh = ext4_bread(handle, inode, *block, EXT4_GET_BLOCKS_CREATE);
 	if (IS_ERR(bh))
Signature · 1 line
2.35.1
author maintainer
#1065 lore raw

PATCH 1063/1095 ext4: remove EA inode entry from mbcache on inode eviction

In reply to Greg Kroah-Hartman

From: Jan Kara <[email protected]>

[ Upstream commit 6bc0d63dad7f9f54d381925ee855b402f652fa39 ]

Currently we remove EA inode from mbcache as soon as its xattr refcount drops to zero. However there can be pending attempts to reuse the inode and thus refcount handling code has to handle the situation when refcount increases from zero anyway. So save some work and just keep EA inode in mbcache until it is getting evicted. At that moment we are sure following iget() of EA inode will fail anyway (or wait for eviction to finish and load things from the disk again) and so removing mbcache entry at that moment is fine and simplifies the code a bit.

---
 fs/ext4/inode.c |  2 ++
 fs/ext4/xattr.c | 24 ++++++++----------------
 fs/ext4/xattr.h |  1 +
 3 files changed, 11 insertions(+), 16 deletions(-)
fs/ext4/inode.c +2 -0
index 9ef6f41a5250..e94ec798dce1 100644
@@ -178,6 +178,8 @@ void ext4_evict_inode(struct inode *inode)
 
 	trace_ext4_evict_inode(inode);
 
+	if (EXT4_I(inode)->i_flags & EXT4_EA_INODE_FL)
+		ext4_evict_ea_inode(inode);
 	if (inode->i_nlink) {
 		/*
 		 * When journalling data dirty buffers are tracked only in the
fs/ext4/xattr.c +8 -16
index c3c3194f3ee1..b57fd07fbdba 100644
@@ -436,6 +436,14 @@ static int ext4_xattr_inode_iget(struct inode *parent, unsigned long ea_ino,
 	return err;
 }
 
+/* Remove entry from mbcache when EA inode is getting evicted */
+void ext4_evict_ea_inode(struct inode *inode)
+{
+	if (EA_INODE_CACHE(inode))
+		mb_cache_entry_delete(EA_INODE_CACHE(inode),
+			ext4_xattr_inode_get_hash(inode), inode->i_ino);
+}
+
 static int
 ext4_xattr_inode_verify_hashes(struct inode *ea_inode,
 			       struct ext4_xattr_entry *entry, void *buffer,
@@ -976,10 +984,8 @@ int __ext4_xattr_set_credits(struct super_block *sb, struct inode *inode,
 static int ext4_xattr_inode_update_ref(handle_t *handle, struct inode *ea_inode,
 				       int ref_change)
 {
-	struct mb_cache *ea_inode_cache = EA_INODE_CACHE(ea_inode);
 	struct ext4_iloc iloc;
 	s64 ref_count;
-	u32 hash;
 	int ret;
 
 	inode_lock(ea_inode);
@@ -1002,14 +1008,6 @@ static int ext4_xattr_inode_update_ref(handle_t *handle, struct inode *ea_inode,
 
 			set_nlink(ea_inode, 1);
 			ext4_orphan_del(handle, ea_inode);
-
-			if (ea_inode_cache) {
-				hash = ext4_xattr_inode_get_hash(ea_inode);
-				mb_cache_entry_create(ea_inode_cache,
-						      GFP_NOFS, hash,
-						      ea_inode->i_ino,
-						      true /* reusable */);
-			}
 		}
 	} else {
 		WARN_ONCE(ref_count < 0, "EA inode %lu ref_count=%lld",
@@ -1022,12 +1020,6 @@ static int ext4_xattr_inode_update_ref(handle_t *handle, struct inode *ea_inode,
 
 			clear_nlink(ea_inode);
 			ext4_orphan_add(handle, ea_inode);
-
-			if (ea_inode_cache) {
-				hash = ext4_xattr_inode_get_hash(ea_inode);
-				mb_cache_entry_delete(ea_inode_cache, hash,
-						      ea_inode->i_ino);
-			}
 		}
 	}
 
fs/ext4/xattr.h +1 -0
index f885f362add4..e5e36bd11f05 100644
@@ -191,6 +191,7 @@ extern void ext4_xattr_inode_array_free(struct ext4_xattr_inode_array *array);
 
 extern int ext4_expand_extra_isize_ea(struct inode *inode, int new_extra_isize,
 			    struct ext4_inode *raw_inode, handle_t *handle);
+extern void ext4_evict_ea_inode(struct inode *inode);
 
 extern const struct xattr_handler *ext4_xattr_handlers[];
 
Signature · 1 line
2.35.1
author maintainer
#1066 lore raw

PATCH 1064/1095 ext4: use kmemdup() to replace kmalloc + memcpy

In reply to Greg Kroah-Hartman

From: Shuqi Zhang <[email protected]>

[ Upstream commit 4efd9f0d120c55b08852ee5605dbb02a77089a5d ]

Replace kmalloc + memcpy with kmemdup()

---
 fs/ext4/xattr.c | 3 +--
 1 file changed, 1 insertion(+), 2 deletions(-)
fs/ext4/xattr.c +1 -2
index b57fd07fbdba..d92d50de5a01 100644
@@ -1887,11 +1887,10 @@ ext4_xattr_block_set(handle_t *handle, struct inode *inode,
 
 			unlock_buffer(bs->bh);
 			ea_bdebug(bs->bh, "cloning");
-			s->base = kmalloc(bs->bh->b_size, GFP_NOFS);
+			s->base = kmemdup(BHDR(bs->bh), bs->bh->b_size, GFP_NOFS);
 			error = -ENOMEM;
 			if (s->base == NULL)
 				goto cleanup;
-			memcpy(s->base, BHDR(bs->bh), bs->bh->b_size);
 			s->first = ENTRY(header(s->base)+1);
 			header(s->base)->h_refcount = cpu_to_le32(1);
 			s->here = ENTRY(s->base + offset);
Signature · 1 line
2.35.1
author maintainer
#1067 lore raw

PATCH 1065/1095 ext4: unindent codeblock in ext4_xattr_block_set()

In reply to Greg Kroah-Hartman

From: Jan Kara <[email protected]>

[ Upstream commit fd48e9acdf26d0cbd80051de07d4a735d05d29b2 ]

Remove unnecessary else (and thus indentation level) from a code block in ext4_xattr_block_set(). It will also make following code changes easier. No functional changes.

---
 fs/ext4/xattr.c | 77 ++++++++++++++++++++++++-------------------------
 1 file changed, 38 insertions(+), 39 deletions(-)
fs/ext4/xattr.c +38 -39
index d92d50de5a01..a25942a74929 100644
@@ -1850,6 +1850,8 @@ ext4_xattr_block_set(handle_t *handle, struct inode *inode,
 #define header(x) ((struct ext4_xattr_header *)(x))
 
 	if (s->base) {
+		int offset = (char *)s->here - bs->bh->b_data;
+
 		BUFFER_TRACE(bs->bh, "get_write_access");
 		error = ext4_journal_get_write_access(handle, sb, bs->bh,
 						      EXT4_JTR_NONE);
@@ -1882,49 +1884,46 @@ ext4_xattr_block_set(handle_t *handle, struct inode *inode,
 			if (error)
 				goto cleanup;
 			goto inserted;
-		} else {
-			int offset = (char *)s->here - bs->bh->b_data;
+		}
+		unlock_buffer(bs->bh);
+		ea_bdebug(bs->bh, "cloning");
+		s->base = kmemdup(BHDR(bs->bh), bs->bh->b_size, GFP_NOFS);
+		error = -ENOMEM;
+		if (s->base == NULL)
+			goto cleanup;
+		s->first = ENTRY(header(s->base)+1);
+		header(s->base)->h_refcount = cpu_to_le32(1);
+		s->here = ENTRY(s->base + offset);
+		s->end = s->base + bs->bh->b_size;
 
-			unlock_buffer(bs->bh);
-			ea_bdebug(bs->bh, "cloning");
-			s->base = kmemdup(BHDR(bs->bh), bs->bh->b_size, GFP_NOFS);
-			error = -ENOMEM;
-			if (s->base == NULL)
+		/*
+		 * If existing entry points to an xattr inode, we need
+		 * to prevent ext4_xattr_set_entry() from decrementing
+		 * ref count on it because the reference belongs to the
+		 * original block. In this case, make the entry look
+		 * like it has an empty value.
+		 */
+		if (!s->not_found && s->here->e_value_inum) {
+			ea_ino = le32_to_cpu(s->here->e_value_inum);
+			error = ext4_xattr_inode_iget(inode, ea_ino,
+				      le32_to_cpu(s->here->e_hash),
+				      &tmp_inode);
+			if (error)
 				goto cleanup;
-			s->first = ENTRY(header(s->base)+1);
-			header(s->base)->h_refcount = cpu_to_le32(1);
-			s->here = ENTRY(s->base + offset);
-			s->end = s->base + bs->bh->b_size;
 
-			/*
-			 * If existing entry points to an xattr inode, we need
-			 * to prevent ext4_xattr_set_entry() from decrementing
-			 * ref count on it because the reference belongs to the
-			 * original block. In this case, make the entry look
-			 * like it has an empty value.
-			 */
-			if (!s->not_found && s->here->e_value_inum) {
-				ea_ino = le32_to_cpu(s->here->e_value_inum);
-				error = ext4_xattr_inode_iget(inode, ea_ino,
-					      le32_to_cpu(s->here->e_hash),
-					      &tmp_inode);
-				if (error)
-					goto cleanup;
-
-				if (!ext4_test_inode_state(tmp_inode,
-						EXT4_STATE_LUSTRE_EA_INODE)) {
-					/*
-					 * Defer quota free call for previous
-					 * inode until success is guaranteed.
-					 */
-					old_ea_inode_quota = le32_to_cpu(
-							s->here->e_value_size);
-				}
-				iput(tmp_inode);
-
-				s->here->e_value_inum = 0;
-				s->here->e_value_size = 0;
+			if (!ext4_test_inode_state(tmp_inode,
+					EXT4_STATE_LUSTRE_EA_INODE)) {
+				/*
+				 * Defer quota free call for previous
+				 * inode until success is guaranteed.
+				 */
+				old_ea_inode_quota = le32_to_cpu(
+						s->here->e_value_size);
 			}
+			iput(tmp_inode);
+
+			s->here->e_value_inum = 0;
+			s->here->e_value_size = 0;
 		}
 	} else {
 		/* Allocate a buffer where we construct the new block. */
Signature · 1 line
2.35.1
author maintainer
#1068 lore raw

PATCH 1066/1095 ext4: fix race when reusing xattr blocks

In reply to Greg Kroah-Hartman

From: Jan Kara <[email protected]>

[ Upstream commit 65f8b80053a1b2fd602daa6814e62d6fa90e5e9b ]

When ext4_xattr_block_set() decides to remove xattr block the following race can happen:

CPU1 CPU2
ext4_xattr_block_set() ext4_xattr_release_block()
new_bh = ext4_xattr_block_cache_find()

                                          lock_buffer(bh);
                                          ref = le32_to_cpu(BHDR(bh)->h_refcount);
                                          if (ref == 1) {
                                            ...
                                            mb_cache_entry_delete();
                                            unlock_buffer(bh);
                                            ext4_free_blocks();
                                              ...
                                              ext4_forget(..., bh, ...);
                                                jbd2_journal_revoke(..., bh);
  ext4_journal_get_write_access(..., new_bh, ...)
    do_get_write_access()
      jbd2_journal_cancel_revoke(..., new_bh);

Later the code in ext4_xattr_block_set() finds out the block got freed and cancels reusal of the block but the revoke stays canceled and so in case of block reuse and journal replay the filesystem can get corrupted. If the race works out slightly differently, we can also hit assertions in the jbd2 code.

Fix the problem by making sure that once matching mbcache entry is found, code dropping the last xattr block reference (or trying to modify xattr block in place) waits until the mbcache entry reference is dropped. This way code trying to reuse xattr block is protected from someone trying to drop the last reference to xattr block.

---
 fs/ext4/xattr.c | 67 +++++++++++++++++++++++++++++++++----------------
 1 file changed, 45 insertions(+), 22 deletions(-)
fs/ext4/xattr.c +45 -22
index a25942a74929..533216e80fa2 100644
@@ -439,9 +439,16 @@ static int ext4_xattr_inode_iget(struct inode *parent, unsigned long ea_ino,
 /* Remove entry from mbcache when EA inode is getting evicted */
 void ext4_evict_ea_inode(struct inode *inode)
 {
-	if (EA_INODE_CACHE(inode))
-		mb_cache_entry_delete(EA_INODE_CACHE(inode),
-			ext4_xattr_inode_get_hash(inode), inode->i_ino);
+	struct mb_cache_entry *oe;
+
+	if (!EA_INODE_CACHE(inode))
+		return;
+	/* Wait for entry to get unused so that we can remove it */
+	while ((oe = mb_cache_entry_delete_or_get(EA_INODE_CACHE(inode),
+			ext4_xattr_inode_get_hash(inode), inode->i_ino))) {
+		mb_cache_entry_wait_unused(oe);
+		mb_cache_entry_put(EA_INODE_CACHE(inode), oe);
+	}
 }
 
 static int
@@ -1229,6 +1236,7 @@ ext4_xattr_release_block(handle_t *handle, struct inode *inode,
 	if (error)
 		goto out;
 
+retry_ref:
 	lock_buffer(bh);
 	hash = le32_to_cpu(BHDR(bh)->h_hash);
 	ref = le32_to_cpu(BHDR(bh)->h_refcount);
@@ -1238,9 +1246,18 @@ ext4_xattr_release_block(handle_t *handle, struct inode *inode,
 		 * This must happen under buffer lock for
 		 * ext4_xattr_block_set() to reliably detect freed block
 		 */
-		if (ea_block_cache)
-			mb_cache_entry_delete(ea_block_cache, hash,
-					      bh->b_blocknr);
+		if (ea_block_cache) {
+			struct mb_cache_entry *oe;
+
+			oe = mb_cache_entry_delete_or_get(ea_block_cache, hash,
+							  bh->b_blocknr);
+			if (oe) {
+				unlock_buffer(bh);
+				mb_cache_entry_wait_unused(oe);
+				mb_cache_entry_put(ea_block_cache, oe);
+				goto retry_ref;
+			}
+		}
 		get_bh(bh);
 		unlock_buffer(bh);
 
@@ -1867,9 +1884,20 @@ ext4_xattr_block_set(handle_t *handle, struct inode *inode,
 			 * ext4_xattr_block_set() to reliably detect modified
 			 * block
 			 */
-			if (ea_block_cache)
-				mb_cache_entry_delete(ea_block_cache, hash,
-						      bs->bh->b_blocknr);
+			if (ea_block_cache) {
+				struct mb_cache_entry *oe;
+
+				oe = mb_cache_entry_delete_or_get(ea_block_cache,
+					hash, bs->bh->b_blocknr);
+				if (oe) {
+					/*
+					 * Xattr block is getting reused. Leave
+					 * it alone.
+					 */
+					mb_cache_entry_put(ea_block_cache, oe);
+					goto clone_block;
+				}
+			}
 			ea_bdebug(bs->bh, "modifying in-place");
 			error = ext4_xattr_set_entry(i, s, handle, inode,
 						     true /* is_block */);
@@ -1885,6 +1913,7 @@ ext4_xattr_block_set(handle_t *handle, struct inode *inode,
 				goto cleanup;
 			goto inserted;
 		}
+clone_block:
 		unlock_buffer(bs->bh);
 		ea_bdebug(bs->bh, "cloning");
 		s->base = kmemdup(BHDR(bs->bh), bs->bh->b_size, GFP_NOFS);
@@ -1990,18 +2019,13 @@ ext4_xattr_block_set(handle_t *handle, struct inode *inode,
 				lock_buffer(new_bh);
 				/*
 				 * We have to be careful about races with
-				 * freeing, rehashing or adding references to
-				 * xattr block. Once we hold buffer lock xattr
-				 * block's state is stable so we can check
-				 * whether the block got freed / rehashed or
-				 * not.  Since we unhash mbcache entry under
-				 * buffer lock when freeing / rehashing xattr
-				 * block, checking whether entry is still
-				 * hashed is reliable. Same rules hold for
-				 * e_reusable handling.
+				 * adding references to xattr block. Once we
+				 * hold buffer lock xattr block's state is
+				 * stable so we can check the additional
+				 * reference fits.
 				 */
-				if (hlist_bl_unhashed(&ce->e_hash_list) ||
-				    !ce->e_reusable) {
+				ref = le32_to_cpu(BHDR(new_bh)->h_refcount) + 1;
+				if (ref > EXT4_XATTR_REFCOUNT_MAX) {
 					/*
 					 * Undo everything and check mbcache
 					 * again.
@@ -2016,9 +2040,8 @@ ext4_xattr_block_set(handle_t *handle, struct inode *inode,
 					new_bh = NULL;
 					goto inserted;
 				}
-				ref = le32_to_cpu(BHDR(new_bh)->h_refcount) + 1;
 				BHDR(new_bh)->h_refcount = cpu_to_le32(ref);
-				if (ref >= EXT4_XATTR_REFCOUNT_MAX)
+				if (ref == EXT4_XATTR_REFCOUNT_MAX)
 					ce->e_reusable = 0;
 				ea_bdebug(new_bh, "reusing; refcount now=%d",
 					  ref);
Signature · 1 line
2.35.1
author maintainer
#1069 lore raw

PATCH 1067/1095 KEYS: asymmetric: enforce SM2 signature use pkey algo

In reply to Greg Kroah-Hartman

From: Tianjia Zhang <[email protected]>

[ Upstream commit 0815291a8fd66cdcf7db1445d4d99b0d16065829 ]

The signature verification of SM2 needs to add the Za value and recalculate sig->digest, which requires the detection of the pkey_algo in public_key_verify_signature(). As Eric Biggers said, the pkey_algo field in sig is attacker-controlled and should be use pkey->pkey_algo instead of sig->pkey_algo, and secondly, if sig->pkey_algo is NULL, it will also cause signature verification failure.

The software_key_determine_akcipher() already forces the algorithms are matched, so the SM3 algorithm is enforced in the SM2 signature, although this has been checked, we still avoid using any algorithm information in the signature as input.

Fixes 215525639631 ("X.509: support OSCCA SM2-with-SM3 certificate verification")
Cc [email protected] # v5.10+
---
 crypto/asymmetric_keys/public_key.c | 7 +++++--
 1 file changed, 5 insertions(+), 2 deletions(-)
crypto/asymmetric_keys/public_key.c +5 -2
index 7c9e6be35c30..2f8352e88860 100644
@@ -304,6 +304,10 @@ static int cert_sig_digest_update(const struct public_key_signature *sig,
 
 	BUG_ON(!sig->data);
 
+	/* SM2 signatures always use the SM3 hash algorithm */
+	if (!sig->hash_algo || strcmp(sig->hash_algo, "sm3") != 0)
+		return -EINVAL;
+
 	ret = sm2_compute_z_digest(tfm_pkey, SM2_DEFAULT_USERID,
 					SM2_DEFAULT_USERID_LEN, dgst);
 	if (ret)
@@ -414,8 +418,7 @@ int public_key_verify_signature(const struct public_key *pkey,
 	if (ret)
 		goto error_free_key;
 
-	if (sig->pkey_algo && strcmp(sig->pkey_algo, "sm2") == 0 &&
-	    sig->data_size) {
+	if (strcmp(pkey->pkey_algo, "sm2") == 0 && sig->data_size) {
 		ret = cert_sig_digest_update(sig, tfm);
 		if (ret)
 			goto error_free_key;
Signature · 1 line
2.35.1
author maintainer
#1070 lore raw

PATCH 1068/1095 tpm: eventlog: Fix section mismatch for DEBUG_SECTION_MISMATCH

In reply to Greg Kroah-Hartman

From: Huacai Chen <[email protected]>

[ Upstream commit bed4593645366ad7362a3aa7bc0d100d8d8236a8 ]

If DEBUG_SECTION_MISMATCH enabled, __calc_tpm2_event_size() will not be inlined, this cause section mismatch like this:

WARNING: modpost: vmlinux.o(.text.unlikely+0xe30c): Section mismatch in reference from the variable L0 to the function .init.text:early_ioremap() The function L0() references
the function __init early_memremap().
This is often because L0 lacks a __init
annotation or the annotation of early_ioremap is wrong.

Fix it by using __always_inline instead of inline for the called-once function __calc_tpm2_event_size().

Fixes 44038bc514a2 ("tpm: Abstract crypto agile event size calculations")
---
 include/linux/tpm_eventlog.h | 2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)
include/linux/tpm_eventlog.h +1 -1
index 739ba9a03ec1..20c0ff54b7a0 100644
@@ -157,7 +157,7 @@ struct tcg_algorithm_info {
  * Return: size of the event on success, 0 on failure
  */
 
-static inline int __calc_tpm2_event_size(struct tcg_pcr_event2_head *event,
+static __always_inline int __calc_tpm2_event_size(struct tcg_pcr_event2_head *event,
 					 struct tcg_pcr_event *event_header,
 					 bool do_mapping)
 {
Signature · 1 line
2.35.1
author maintainer
#1071 lore raw

PATCH 1069/1095 tpm: Add check for Failure mode for TPM2 modules

In reply to Greg Kroah-Hartman

From: Mårten Lindahl <[email protected]>

[ Upstream commit 863ed94c589fcd1984f4e3080f069d30508044bb ]

In commit 0aa698787aa2 ("tpm: Add Upgrade/Reduced mode support for TPM2 modules") it was said that:

"If the TPM is in Failure mode, it will successfully respond to both tpm2_do_selftest() and tpm2_startup() calls. Although, will fail to answer to tpm2_get_cc_attrs_tbl(). Use this fact to conclude that TPM is in Failure mode."

But a check was never added in the commit when calling tpm2_get_cc_attrs_tbl() to conclude that the TPM is in Failure mode. This commit corrects this by adding a check.

Fixes 0aa698787aa2 ("tpm: Add Upgrade/Reduced mode support for TPM2 modules")
Cc [email protected] # v5.17+
---
 drivers/char/tpm/tpm2-cmd.c | 6 ++++++
 1 file changed, 6 insertions(+)
drivers/char/tpm/tpm2-cmd.c +6 -0
index 04a3e23a4afc..4419593d9531 100644
@@ -752,6 +752,12 @@ int tpm2_auto_startup(struct tpm_chip *chip)
 	}
 
 	rc = tpm2_get_cc_attrs_tbl(chip);
+	if (rc == TPM2_RC_FAILURE || (rc < 0 && rc != -ENOMEM)) {
+		dev_info(&chip->dev,
+			 "TPM in field failure mode, requires firmware upgrade\n");
+		chip->flags |= TPM_CHIP_FLAG_FIRMWARE_UPGRADE;
+		rc = 0;
+	}
 
 out:
 	if (rc == TPM2_RC_UPGRADE) {
Signature · 1 line
2.35.1
author maintainer
#1072 lore raw

PATCH 1070/1095 xen-blkback: fix persistent grants negotiation

In reply to Greg Kroah-Hartman

From: SeongJae Park <[email protected]>

commit fc9be616bb8f3ed9cf560308f86904f5c06be205 upstream.

Persistent grants feature can be used only when both backend and the frontend supports the feature. The feature was always supported by 'blkback', but commit aac8a70db24b ("xen-blkback: add a parameter for disabling of persistent grants") has introduced a parameter for disabling it runtime.

To avoid the parameter be updated while being used by 'blkback', the commit caches the parameter into 'vbd->feature_gnt_persistent' in 'xen_vbd_create()', and then check if the guest also supports the feature and finally updates the field in 'connect_ring()'.

However, 'connect_ring()' could be called before 'xen_vbd_create()', so later execution of 'xen_vbd_create()' can wrongly overwrite 'true' to 'vbd->feature_gnt_persistent'. As a result, 'blkback' could try to use 'persistent grants' feature even if the guest doesn't support the feature.

This commit fixes the issue by moving the parameter value caching to 'xen_blkif_alloc()', which allocates the 'blkif'. Because the struct embeds 'vbd' object, which will be used by 'connect_ring()' later, this should be called before 'connect_ring()' and therefore this should be the right and safe place to do the caching.

---
 drivers/block/xen-blkback/xenbus.c |   15 +++++++--------
 1 file changed, 7 insertions(+), 8 deletions(-)
drivers/block/xen-blkback/xenbus.c +7 -8
@@ -157,6 +157,11 @@ static int xen_blkif_alloc_rings(struct
 	return 0;
 }
 
+/* Enable the persistent grants feature. */
+static bool feature_persistent = true;
+module_param(feature_persistent, bool, 0644);
+MODULE_PARM_DESC(feature_persistent, "Enables the persistent grants feature");
+
 static struct xen_blkif *xen_blkif_alloc(domid_t domid)
 {
 	struct xen_blkif *blkif;
@@ -181,6 +186,8 @@ static struct xen_blkif *xen_blkif_alloc
 	__module_get(THIS_MODULE);
 	INIT_WORK(&blkif->free_work, xen_blkif_deferred_free);
 
+	blkif->vbd.feature_gnt_persistent = feature_persistent;
+
 	return blkif;
 }
 
@@ -472,12 +479,6 @@ static void xen_vbd_free(struct xen_vbd
 	vbd->bdev = NULL;
 }
 
-/* Enable the persistent grants feature. */
-static bool feature_persistent = true;
-module_param(feature_persistent, bool, 0644);
-MODULE_PARM_DESC(feature_persistent,
-		"Enables the persistent grants feature");
-
 static int xen_vbd_create(struct xen_blkif *blkif, blkif_vdev_t handle,
 			  unsigned major, unsigned minor, int readonly,
 			  int cdrom)
@@ -523,8 +524,6 @@ static int xen_vbd_create(struct xen_blk
 	if (q && blk_queue_secure_erase(q))
 		vbd->discard_secure = true;
 
-	vbd->feature_gnt_persistent = feature_persistent;
-
 	pr_debug("Successful creation of handle=%04x (dom=%u)\n",
 		handle, blkif->domid);
 	return 0;
author maintainer
#1073 lore raw

PATCH 1071/1095 xen-blkback: Apply feature_persistent parameter when connect

In reply to Greg Kroah-Hartman

From: Maximilian Heyne <[email protected]>

commit e94c6101e151b019b8babc518ac2a6ada644a5a1 upstream.

In some use cases[1], the backend is created while the frontend doesn't support the persistent grants feature, but later the frontend can be changed to support the feature and reconnect. In the past, 'blkback' enabled the persistent grants feature since it unconditionally checked if frontend supports the persistent grants feature for every connect ('connect_ring()') and decided whether it should use persistent grans or not.

However, commit aac8a70db24b ("xen-blkback: add a parameter for disabling of persistent grants") has mistakenly changed the behavior. It made the frontend feature support check to not be repeated once it shown the 'feature_persistent' as 'false', or the frontend doesn't support persistent grants.

This commit changes the behavior of the parameter to make effect for every connect, so that the previous workflow can work again as expected.

[1] https://lore.kernel.org/xen-devel/CAJwUmVB6H3iTs-C+U=v-pwJB7-_ZRHPxHzKRJZ22xEPW7z8a=g@mail.gmail.com/

---
 Documentation/ABI/testing/sysfs-driver-xen-blkback |    2 +-
 drivers/block/xen-blkback/xenbus.c                 |    9 +++------
 2 files changed, 4 insertions(+), 7 deletions(-)
Documentation/ABI/testing/sysfs-driver-xen-blkback +1 -1
@@ -42,5 +42,5 @@ KernelVersion:  5.10
 Contact:        SeongJae Park <[email protected]>
 Description:
                 Whether to enable the persistent grants feature or not.  Note
-                that this option only takes effect on newly created backends.
+                that this option only takes effect on newly connected backends.
                 The default is Y (enable).
drivers/block/xen-blkback/xenbus.c +3 -6
@@ -186,8 +186,6 @@ static struct xen_blkif *xen_blkif_alloc
 	__module_get(THIS_MODULE);
 	INIT_WORK(&blkif->free_work, xen_blkif_deferred_free);
 
-	blkif->vbd.feature_gnt_persistent = feature_persistent;
-
 	return blkif;
 }
 
@@ -1090,10 +1088,9 @@ static int connect_ring(struct backend_i
 		xenbus_dev_fatal(dev, err, "unknown fe protocol %s", protocol);
 		return -ENOSYS;
 	}
-	if (blkif->vbd.feature_gnt_persistent)
-		blkif->vbd.feature_gnt_persistent =
-			xenbus_read_unsigned(dev->otherend,
-					"feature-persistent", 0);
+
+	blkif->vbd.feature_gnt_persistent = feature_persistent &&
+		xenbus_read_unsigned(dev->otherend, "feature-persistent", 0);
 
 	blkif->vbd.overflow_max_grants = 0;
 
author maintainer
#1074 lore raw

PATCH 1072/1095 xen-blkfront: Apply feature_persistent parameter when connect

In reply to Greg Kroah-Hartman

From: SeongJae Park <[email protected]>

commit 402c43ea6b34a1b371ffeed9adf907402569eaf5 upstream.

In some use cases[1], the backend is created while the frontend doesn't support the persistent grants feature, but later the frontend can be changed to support the feature and reconnect. In the past, 'blkback' enabled the persistent grants feature since it unconditionally checked if frontend supports the persistent grants feature for every connect ('connect_ring()') and decided whether it should use persistent grans or not.

However, commit aac8a70db24b ("xen-blkback: add a parameter for disabling of persistent grants") has mistakenly changed the behavior. It made the frontend feature support check to not be repeated once it shown the 'feature_persistent' as 'false', or the frontend doesn't support persistent grants.

Similar behavioral change has made on 'blkfront' by commit 74a852479c68 ("xen-blkfront: add a parameter for disabling of persistent grants"). This commit changes the behavior of the parameter to make effect for every connect, so that the previous behavior of 'blkfront' can be restored.

[1] https://lore.kernel.org/xen-devel/CAJwUmVB6H3iTs-C+U=v-pwJB7-_ZRHPxHzKRJZ22xEPW7z8a=g@mail.gmail.com/

Fixes 74a852479c68 ("xen-blkfront: add a parameter for disabling of persistent grants")
Cc <[email protected]> # 5.10.x
---
 Documentation/ABI/testing/sysfs-driver-xen-blkfront |    2 +-
 drivers/block/xen-blkfront.c                        |    4 +---
 2 files changed, 2 insertions(+), 4 deletions(-)
Documentation/ABI/testing/sysfs-driver-xen-blkfront +1 -1
@@ -15,5 +15,5 @@ KernelVersion:  5.10
 Contact:        SeongJae Park <[email protected]>
 Description:
                 Whether to enable the persistent grants feature or not.  Note
-                that this option only takes effect on newly created frontends.
+                that this option only takes effect on newly connected frontends.
                 The default is Y (enable).
drivers/block/xen-blkfront.c +1 -3
@@ -2011,8 +2011,6 @@ static int blkfront_probe(struct xenbus_
 	info->vdevice = vdevice;
 	info->connected = BLKIF_STATE_DISCONNECTED;
 
-	info->feature_persistent = feature_persistent;
-
 	/* Front end dir is a number, which is used as the id. */
 	info->handle = simple_strtoul(strrchr(dev->nodename, '/')+1, NULL, 0);
 	dev_set_drvdata(&dev->dev, info);
@@ -2306,7 +2304,7 @@ static void blkfront_gather_backend_feat
 	if (xenbus_read_unsigned(info->xbdev->otherend, "feature-discard", 0))
 		blkfront_setup_discard(info);
 
-	if (info->feature_persistent)
+	if (feature_persistent)
 		info->feature_persistent =
 			!!xenbus_read_unsigned(info->xbdev->otherend,
 					       "feature-persistent", 0);
author maintainer
#1075 lore raw

PATCH 1073/1095 powerpc: Fix eh field when calling lwarx on PPC32

In reply to Greg Kroah-Hartman

From: Christophe Leroy <[email protected]>

commit 18db466a9a306406dab3b134014d9f6ed642471c upstream.

Commit 9401f4e46cf6 ("powerpc: Use lwarx/ldarx directly instead of PPC_LWARX/LDARX macros") properly handled the eh field of lwarx in asm/bitops.h but failed to clear it for PPC32 in asm/simple_spinlock.h

So, do as in arch_atomic_try_cmpxchg_lock(), set it to 1 if PPC64 but set it to 0 if PPC32. For that use IS_ENABLED(CONFIG_PPC64) which returns 1 when CONFIG_PPC64 is set and 0 otherwise.

Fixes 9401f4e46cf6 ("powerpc: Use lwarx/ldarx directly instead of PPC_LWARX/LDARX macros")
Cc [email protected] # v5.15+
[mpe: Use symbolic names, use 'n' constraint per Segher]
---
 arch/powerpc/include/asm/simple_spinlock.h |   15 +++++++++------
 1 file changed, 9 insertions(+), 6 deletions(-)
arch/powerpc/include/asm/simple_spinlock.h +9 -6
@@ -48,10 +48,11 @@ static inline int arch_spin_is_locked(ar
 static inline unsigned long __arch_spin_trylock(arch_spinlock_t *lock)
 {
 	unsigned long tmp, token;
+	unsigned int eh = IS_ENABLED(CONFIG_PPC64);
 
 	token = LOCK_TOKEN;
 	__asm__ __volatile__(
-"1:	lwarx		%0,0,%2,1\n\
+"1:	lwarx		%0,0,%2,%[eh]\n\
 	cmpwi		0,%0,0\n\
 	bne-		2f\n\
 	stwcx.		%1,0,%2\n\
@@ -59,7 +60,7 @@ static inline unsigned long __arch_spin_
 	PPC_ACQUIRE_BARRIER
 "2:"
 	: "=&r" (tmp)
-	: "r" (token), "r" (&lock->slock)
+	: "r" (token), "r" (&lock->slock), [eh] "n" (eh)
 	: "cr0", "memory");
 
 	return tmp;
@@ -156,9 +157,10 @@ static inline void arch_spin_unlock(arch
 static inline long __arch_read_trylock(arch_rwlock_t *rw)
 {
 	long tmp;
+	unsigned int eh = IS_ENABLED(CONFIG_PPC64);
 
 	__asm__ __volatile__(
-"1:	lwarx		%0,0,%1,1\n"
+"1:	lwarx		%0,0,%1,%[eh]\n"
 	__DO_SIGN_EXTEND
 "	addic.		%0,%0,1\n\
 	ble-		2f\n"
@@ -166,7 +168,7 @@ static inline long __arch_read_trylock(a
 	bne-		1b\n"
 	PPC_ACQUIRE_BARRIER
 "2:"	: "=&r" (tmp)
-	: "r" (&rw->lock)
+	: "r" (&rw->lock), [eh] "n" (eh)
 	: "cr0", "xer", "memory");
 
 	return tmp;
@@ -179,17 +181,18 @@ static inline long __arch_read_trylock(a
 static inline long __arch_write_trylock(arch_rwlock_t *rw)
 {
 	long tmp, token;
+	unsigned int eh = IS_ENABLED(CONFIG_PPC64);
 
 	token = WRLOCK_TOKEN;
 	__asm__ __volatile__(
-"1:	lwarx		%0,0,%2,1\n\
+"1:	lwarx		%0,0,%2,%[eh]\n\
 	cmpwi		0,%0,0\n\
 	bne-		2f\n"
 "	stwcx.		%1,0,%2\n\
 	bne-		1b\n"
 	PPC_ACQUIRE_BARRIER
 "2:"	: "=&r" (tmp)
-	: "r" (token), "r" (&rw->lock)
+	: "r" (token), "r" (&rw->lock), [eh] "n" (eh)
 	: "cr0", "memory");
 
 	return tmp;
author maintainer
#1076 lore raw

PATCH 1074/1095 btrfs: join running log transaction when logging new name

In reply to Greg Kroah-Hartman

From: Filipe Manana <[email protected]>

commit 723df2bcc9e166ac7fb82b3932a53e09415dfcde upstream.

When logging a new name, in case of a rename, we pin the log before changing it. We then either delete a directory entry from the log or insert a key range item to mark the old name for deletion on log replay.

However when doing one of those log changes we may have another task that started writing out the log (at btrfs_sync_log()) and it started before we pinned the log root. So we may end up changing a log tree while its writeback is being started by another task syncing the log. This can lead to inconsistencies in a log tree and other unexpected results during log replay, because we can get some committed node pointing to a node/leaf that ends up not getting written to disk before the next log commit.

The problem, conceptually, started to happen in commit 88d2beec7e53fc ("btrfs: avoid logging all directory changes during renames"), because there we started to update the log without joining its current transaction first.

However the problem only became visible with commit 259c4b96d78dda ("btrfs: stop doing unnecessary log updates during a rename"), and that is because we used to pin the log at btrfs_rename() and then before entering btrfs_log_new_name(), when unlinking the old dentry, we ended up at btrfs_del_inode_ref_in_log() and btrfs_del_dir_entries_in_log(). Both of them join the current log transaction, effectively waiting for any log transaction writeout (due to acquiring the root's log_mutex). This made it safe even after leaving the current log transaction, because we remained with the log pinned when we called btrfs_log_new_name().

Then in commit 259c4b96d78dda ("btrfs: stop doing unnecessary log updates during a rename"), we removed the log pinning from btrfs_rename() and stopped calling btrfs_del_inode_ref_in_log() and btrfs_del_dir_entries_in_log() during the rename, and started to do all the needed work at btrfs_log_new_name(), but without joining the current log transaction, only pinning the log, which is racy because another task may have started writeout of the log tree right before we pinned the log.

Both commits landed in kernel 5.18, so it doesn't make any practical difference which should be blamed, but I'm blaming the second commit only because with the first one, by chance, the problem did not happen due to the fact we joined the log transaction after pinning the log and unpinned it only after calling btrfs_log_new_name().

So make btrfs_log_new_name() join the current log transaction instead of pinning it, so that we never do log updates if it's writeout is starting.

Fixes 259c4b96d78dda ("btrfs: stop doing unnecessary log updates during a rename")
---
 fs/btrfs/tree-log.c |    9 ++++++++-
 1 file changed, 8 insertions(+), 1 deletion(-)
fs/btrfs/tree-log.c +8 -1
@@ -7030,8 +7030,15 @@ void btrfs_log_new_name(struct btrfs_tra
 		 * anyone from syncing the log until we have updated both inodes
 		 * in the log.
 		 */
+		ret = join_running_log_trans(root);
+		/*
+		 * At least one of the inodes was logged before, so this should
+		 * not fail, but if it does, it's not serious, just bail out and
+		 * mark the log for a full commit.
+		 */
+		if (WARN_ON_ONCE(ret < 0))
+			goto out;
 		log_pinned = true;
-		btrfs_pin_log_trans(root);
 
 		path = btrfs_alloc_path();
 		if (!path) {
author maintainer
#1077 lore raw

PATCH 1075/1095 btrfs: convert count_max_extents() to use fs_info->max_extent_size

In reply to Greg Kroah-Hartman

From: Naohiro Aota <[email protected]>

commit 7d7672bc5d1038c745716c397d892d21e29de71c upstream.

If count_max_extents() uses BTRFS_MAX_EXTENT_SIZE to calculate the number of extents needed, btrfs release the metadata reservation too much on its way to write out the data.

Now that BTRFS_MAX_EXTENT_SIZE is replaced with fs_info->max_extent_size, convert count_max_extents() to use it instead, and fix the calculation of the metadata reservation.

Fixes d8e3fb106f39 ("btrfs: zoned: use ZONE_APPEND write for zoned mode")
---
 fs/btrfs/ctree.h          |   21 +++++++++++++--------
 fs/btrfs/delalloc-space.c |    6 +++---
 fs/btrfs/inode.c          |   16 ++++++++--------
 3 files changed, 24 insertions(+), 19 deletions(-)
fs/btrfs/ctree.h +13 -8
@@ -107,14 +107,6 @@ struct btrfs_ioctl_encoded_io_args;
 #define BTRFS_STAT_CURR		0
 #define BTRFS_STAT_PREV		1
 
-/*
- * Count how many BTRFS_MAX_EXTENT_SIZE cover the @size
- */
-static inline u32 count_max_extents(u64 size)
-{
-	return div_u64(size + BTRFS_MAX_EXTENT_SIZE - 1, BTRFS_MAX_EXTENT_SIZE);
-}
-
 static inline unsigned long btrfs_chunk_item_size(int num_stripes)
 {
 	BUG_ON(num_stripes == 0);
@@ -3945,6 +3937,19 @@ static inline bool btrfs_is_zoned(const
 	return fs_info->zoned != 0;
 }
 
+/*
+ * Count how many fs_info->max_extent_size cover the @size
+ */
+static inline u32 count_max_extents(struct btrfs_fs_info *fs_info, u64 size)
+{
+#ifdef CONFIG_BTRFS_FS_RUN_SANITY_TESTS
+	if (!fs_info)
+		return div_u64(size + BTRFS_MAX_EXTENT_SIZE - 1, BTRFS_MAX_EXTENT_SIZE);
+#endif
+
+	return div_u64(size + fs_info->max_extent_size - 1, fs_info->max_extent_size);
+}
+
 static inline bool btrfs_is_data_reloc_root(const struct btrfs_root *root)
 {
 	return root->root_key.objectid == BTRFS_DATA_RELOC_TREE_OBJECTID;
fs/btrfs/delalloc-space.c +3 -3
@@ -273,7 +273,7 @@ static void calc_inode_reservations(stru
 				    u64 num_bytes, u64 disk_num_bytes,
 				    u64 *meta_reserve, u64 *qgroup_reserve)
 {
-	u64 nr_extents = count_max_extents(num_bytes);
+	u64 nr_extents = count_max_extents(fs_info, num_bytes);
 	u64 csum_leaves = btrfs_csum_bytes_to_leaves(fs_info, disk_num_bytes);
 	u64 inode_update = btrfs_calc_metadata_size(fs_info, 1);
 
@@ -349,7 +349,7 @@ int btrfs_delalloc_reserve_metadata(stru
 	 * needs to free the reservation we just made.
 	 */
 	spin_lock(&inode->lock);
-	nr_extents = count_max_extents(num_bytes);
+	nr_extents = count_max_extents(fs_info, num_bytes);
 	btrfs_mod_outstanding_extents(inode, nr_extents);
 	inode->csum_bytes += disk_num_bytes;
 	btrfs_calculate_inode_block_rsv_size(fs_info, inode);
@@ -412,7 +412,7 @@ void btrfs_delalloc_release_extents(stru
 	unsigned num_extents;
 
 	spin_lock(&inode->lock);
-	num_extents = count_max_extents(num_bytes);
+	num_extents = count_max_extents(fs_info, num_bytes);
 	btrfs_mod_outstanding_extents(inode, -num_extents);
 	btrfs_calculate_inode_block_rsv_size(fs_info, inode);
 	spin_unlock(&inode->lock);
fs/btrfs/inode.c +8 -8
@@ -2159,10 +2159,10 @@ void btrfs_split_delalloc_extent(struct
 		 * applies here, just in reverse.
 		 */
 		new_size = orig->end - split + 1;
-		num_extents = count_max_extents(new_size);
+		num_extents = count_max_extents(fs_info, new_size);
 		new_size = split - orig->start;
-		num_extents += count_max_extents(new_size);
-		if (count_max_extents(size) >= num_extents)
+		num_extents += count_max_extents(fs_info, new_size);
+		if (count_max_extents(fs_info, size) >= num_extents)
 			return;
 	}
 
@@ -2219,10 +2219,10 @@ void btrfs_merge_delalloc_extent(struct
 	 * this case.
 	 */
 	old_size = other->end - other->start + 1;
-	num_extents = count_max_extents(old_size);
+	num_extents = count_max_extents(fs_info, old_size);
 	old_size = new->end - new->start + 1;
-	num_extents += count_max_extents(old_size);
-	if (count_max_extents(new_size) >= num_extents)
+	num_extents += count_max_extents(fs_info, old_size);
+	if (count_max_extents(fs_info, new_size) >= num_extents)
 		return;
 
 	spin_lock(&BTRFS_I(inode)->lock);
@@ -2301,7 +2301,7 @@ void btrfs_set_delalloc_extent(struct in
 	if (!(state->state & EXTENT_DELALLOC) && (*bits & EXTENT_DELALLOC)) {
 		struct btrfs_root *root = BTRFS_I(inode)->root;
 		u64 len = state->end + 1 - state->start;
-		u32 num_extents = count_max_extents(len);
+		u32 num_extents = count_max_extents(fs_info, len);
 		bool do_list = !btrfs_is_free_space_inode(BTRFS_I(inode));
 
 		spin_lock(&BTRFS_I(inode)->lock);
@@ -2343,7 +2343,7 @@ void btrfs_clear_delalloc_extent(struct
 	struct btrfs_inode *inode = BTRFS_I(vfs_inode);
 	struct btrfs_fs_info *fs_info = btrfs_sb(vfs_inode->i_sb);
 	u64 len = state->end + 1 - state->start;
-	u32 num_extents = count_max_extents(len);
+	u32 num_extents = count_max_extents(fs_info, len);
 
 	if ((state->state & EXTENT_DEFRAG) && (*bits & EXTENT_DEFRAG)) {
 		spin_lock(&inode->lock);
author maintainer
#1078 lore raw

PATCH 1076/1095 net_sched: cls_route: remove from list when handle is 0

In reply to Greg Kroah-Hartman

From: Thadeu Lima de Souza Cascardo <[email protected]>

commit 9ad36309e2719a884f946678e0296be10f0bb4c1 upstream.

When a route filter is replaced and the old filter has a 0 handle, the old one won't be removed from the hashtable, while it will still be freed.

The test was there since before commit 1109c00547fc ("net: sched: RCU cls_route"), when a new filter was not allocated when there was an old one. The old filter was reused and the reinserting would only be necessary if an old filter was replaced. That was still wrong for the same case where the old handle was 0.

Remove the old filter from the list independently from its handle value.

This fixes CVE-2022-2588, also reported as ZDI-CAN-17440.

---
 net/sched/cls_route.c |    2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)
net/sched/cls_route.c +1 -1
@@ -526,7 +526,7 @@ static int route4_change(struct net *net
 	rcu_assign_pointer(f->next, f1);
 	rcu_assign_pointer(*fp, f);
 
-	if (fold && fold->handle && f->handle != fold->handle) {
+	if (fold) {
 		th = to_hash(fold->handle);
 		h = from_hash(fold->handle >> 16);
 		b = rtnl_dereference(head->table[th]);
author maintainer
#1079 lore raw

PATCH 1077/1095 tcp: fix over estimation in sk_forced_mem_schedule()

In reply to Greg Kroah-Hartman

From: Eric Dumazet <[email protected]>

commit c4ee118561a0f74442439b7b5b486db1ac1ddfeb upstream.

sk_forced_mem_schedule() has a bug similar to ones fixed in commit 7c80b038d23e ("net: fix sk_wmem_schedule() and sk_rmem_schedule() errors")

While this bug has little chance to trigger in old kernels, we need to fix it before the following patch.

Fixes d83769a580f1 ("tcp: fix possible deadlock in tcp_send_fin()")
---
 net/ipv4/tcp_output.c |    7 ++++---
 1 file changed, 4 insertions(+), 3 deletions(-)
net/ipv4/tcp_output.c +4 -3
@@ -3367,11 +3367,12 @@ void tcp_xmit_retransmit_queue(struct so
  */
 void sk_forced_mem_schedule(struct sock *sk, int size)
 {
-	int amt;
+	int delta, amt;
 
-	if (size <= sk->sk_forward_alloc)
+	delta = size - sk->sk_forward_alloc;
+	if (delta <= 0)
 		return;
-	amt = sk_mem_pages(size);
+	amt = sk_mem_pages(delta);
 	sk->sk_forward_alloc += amt * SK_MEM_QUANTUM;
 	sk_memory_allocated_add(sk, amt);
 
author maintainer
#1080 lore raw

PATCH 1078/1095 crypto: lib/blake2s - reduce stack frame usage in self test

In reply to Greg Kroah-Hartman

From: Jason A. Donenfeld <[email protected]>

commit d6c14da474bf260d73953fbf7992c98d9112aec7 upstream.

Using 3 blocks here doesn't give us much more than using 2, and it causes a stack frame size warning on certain compiler/config/arch combinations:

lib/crypto/blake2s-selftest.c: In function 'blake2s_selftest':

Quoted 1 line
Quoted 1 line

lib/crypto/blake2s-selftest.c:632:1: warning: the frame size of 1088 bytes is larger than 1024 bytes [-Wframe-larger-than=]

     632 | }
         | ^

So this patch just reduces the block from 3 to 2, which makes the warning go away.

---
 lib/crypto/blake2s-selftest.c |    6 +++---
 1 file changed, 3 insertions(+), 3 deletions(-)
lib/crypto/blake2s-selftest.c +3 -3
@@ -593,7 +593,7 @@ bool __init blake2s_selftest(void)
 		enum { TEST_ALIGNMENT = 16 };
 		u8 unaligned_block[BLAKE2S_BLOCK_SIZE + TEST_ALIGNMENT - 1]
 					__aligned(TEST_ALIGNMENT);
-		u8 blocks[BLAKE2S_BLOCK_SIZE * 3];
+		u8 blocks[BLAKE2S_BLOCK_SIZE * 2];
 		struct blake2s_state state1, state2;
 
 		get_random_bytes(blocks, sizeof(blocks));
@@ -603,8 +603,8 @@ bool __init blake2s_selftest(void)
     defined(CONFIG_CRYPTO_ARCH_HAVE_LIB_BLAKE2S)
 		memcpy(&state1, &state, sizeof(state1));
 		memcpy(&state2, &state, sizeof(state2));
-		blake2s_compress(&state1, blocks, 3, BLAKE2S_BLOCK_SIZE);
-		blake2s_compress_generic(&state2, blocks, 3, BLAKE2S_BLOCK_SIZE);
+		blake2s_compress(&state1, blocks, 2, BLAKE2S_BLOCK_SIZE);
+		blake2s_compress_generic(&state2, blocks, 2, BLAKE2S_BLOCK_SIZE);
 		if (memcmp(&state1, &state2, sizeof(state1))) {
 			pr_err("blake2s random compress self-test %d: FAIL\n",
 			       i + 1);
author maintainer
#1081 lore raw

PATCH 1079/1095 raw: remove unused variables from raw6_icmp_error()

In reply to Greg Kroah-Hartman

From: Eric Dumazet <[email protected]>

commit c4fceb46add65481ef0dfb79cad24c3c269b4cad upstream.

saddr and daddr are set but not used.

---
 net/ipv6/raw.c |    3 ---
 1 file changed, 3 deletions(-)
net/ipv6/raw.c +0 -3
@@ -332,7 +332,6 @@ static void rawv6_err(struct sock *sk, s
 void raw6_icmp_error(struct sk_buff *skb, int nexthdr,
 		u8 type, u8 code, int inner_offset, __be32 info)
 {
-	const struct in6_addr *saddr, *daddr;
 	struct net *net = dev_net(skb->dev);
 	struct hlist_nulls_head *hlist;
 	struct hlist_nulls_node *hnode;
@@ -345,8 +344,6 @@ void raw6_icmp_error(struct sk_buff *skb
 	hlist_nulls_for_each_entry(sk, hnode, hlist, sk_nulls_node) {
 		/* Note: ipv6_hdr(skb) != skb->data */
 		const struct ipv6hdr *ip6h = (const struct ipv6hdr *)skb->data;
-		saddr = &ip6h->saddr;
-		daddr = &ip6h->daddr;
 
 		if (!raw_v6_match(net, sk, nexthdr, &ip6h->saddr, &ip6h->daddr,
 				  inet6_iif(skb), inet6_iif(skb)))
author maintainer
#1082 lore raw

PATCH 1080/1095 raw: fix a typo in raw_icmp_error()

In reply to Greg Kroah-Hartman

From: Eric Dumazet <[email protected]>

commit 97a4d46b1516250d640c1ae0c9e7129d160d6a1c upstream.

I accidentally broke IPv4 traceroute, by swapping iph->saddr and iph->daddr.

Probably because raw_icmp_error() and raw_v4_input() use different order for iph->saddr and iph->daddr.

---
 net/ipv4/raw.c |    2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)
net/ipv4/raw.c +1 -1
@@ -278,7 +278,7 @@ void raw_icmp_error(struct sk_buff *skb,
 	hlist_nulls_for_each_entry(sk, hnode, hlist, sk_nulls_node) {
 		iph = (const struct iphdr *)skb->data;
 		if (!raw_v4_match(net, sk, iph->protocol,
-				  iph->saddr, iph->daddr, dif, sdif))
+				  iph->daddr, iph->saddr, dif, sdif))
 			continue;
 		raw_err(sk, skb, info);
 	}
author maintainer
#1083 lore raw

PATCH 1081/1095 Revert "mwifiex: fix sleep in atomic context bugs caused by dev_coredumpv"

In reply to Greg Kroah-Hartman

From: Greg Kroah-Hartman <[email protected]>

commit 5f8954e099b8ae96e7de1bb95950e00c85bedd40 upstream.

This reverts commit a52ed4866d2b90dd5e4ae9dabd453f3ed8fa3cbc as it causes build problems in linux-next. It needs to be reintroduced in a way that can allow the api to evolve and not require a "flag day" to catch all users.

---
 drivers/net/wireless/marvell/mwifiex/init.c      |    9 ++++-----
 drivers/net/wireless/marvell/mwifiex/main.h      |    3 +--
 drivers/net/wireless/marvell/mwifiex/sta_event.c |    6 +++---
 3 files changed, 8 insertions(+), 10 deletions(-)
drivers/net/wireless/marvell/mwifiex/init.c +4 -5
@@ -63,10 +63,9 @@ static void wakeup_timer_fn(struct timer
 		adapter->if_ops.card_reset(adapter);
 }
 
-static void fw_dump_work(struct work_struct *work)
+static void fw_dump_timer_fn(struct timer_list *t)
 {
-	struct mwifiex_adapter *adapter =
-		container_of(work, struct mwifiex_adapter, devdump_work.work);
+	struct mwifiex_adapter *adapter = from_timer(adapter, t, devdump_timer);
 
 	mwifiex_upload_device_dump(adapter);
 }
@@ -322,7 +321,7 @@ static void mwifiex_init_adapter(struct
 	adapter->active_scan_triggered = false;
 	timer_setup(&adapter->wakeup_timer, wakeup_timer_fn, 0);
 	adapter->devdump_len = 0;
-	INIT_DELAYED_WORK(&adapter->devdump_work, fw_dump_work);
+	timer_setup(&adapter->devdump_timer, fw_dump_timer_fn, 0);
 }
 
 /*
@@ -401,7 +400,7 @@ static void
 mwifiex_adapter_cleanup(struct mwifiex_adapter *adapter)
 {
 	del_timer(&adapter->wakeup_timer);
-	cancel_delayed_work_sync(&adapter->devdump_work);
+	del_timer_sync(&adapter->devdump_timer);
 	mwifiex_cancel_all_pending_cmd(adapter);
 	wake_up_interruptible(&adapter->cmd_wait_q.wait);
 	wake_up_interruptible(&adapter->hs_activate_wait_q);
drivers/net/wireless/marvell/mwifiex/main.h +1 -2
@@ -49,7 +49,6 @@
 #include <linux/pm_runtime.h>
 #include <linux/slab.h>
 #include <linux/of_irq.h>
-#include <linux/workqueue.h>
 
 #include "decl.h"
 #include "ioctl.h"
@@ -1056,7 +1055,7 @@ struct mwifiex_adapter {
 	/* Device dump data/length */
 	void *devdump_data;
 	int devdump_len;
-	struct delayed_work devdump_work;
+	struct timer_list devdump_timer;
 
 	bool ignore_btcoex_events;
 };
drivers/net/wireless/marvell/mwifiex/sta_event.c +3 -3
@@ -623,8 +623,8 @@ mwifiex_fw_dump_info_event(struct mwifie
 		 * transmission event get lost, in this cornel case,
 		 * user would still get partial of the dump.
 		 */
-		schedule_delayed_work(&adapter->devdump_work,
-				      msecs_to_jiffies(MWIFIEX_TIMER_10S));
+		mod_timer(&adapter->devdump_timer,
+			  jiffies + msecs_to_jiffies(MWIFIEX_TIMER_10S));
 	}
 
 	/* Overflow check */
@@ -643,7 +643,7 @@ mwifiex_fw_dump_info_event(struct mwifie
 	return;
 
 upload_dump:
-	cancel_delayed_work_sync(&adapter->devdump_work);
+	del_timer_sync(&adapter->devdump_timer);
 	mwifiex_upload_device_dump(adapter);
 }
 
author maintainer
#1084 lore raw

PATCH 1082/1095 mptcp: refine memory scheduling

In reply to Greg Kroah-Hartman

From: Paolo Abeni <[email protected]>

commit 69d93daec026cdda98e29e8edb12534bfa5b1a9b upstream.

Similar to commit 7c80b038d23e ("net: fix sk_wmem_schedule() and sk_rmem_schedule() errors"), let the MPTCP receive path schedule exactly the required amount of memory.

---
 net/mptcp/protocol.c |    3 ++-
 1 file changed, 2 insertions(+), 1 deletion(-)
net/mptcp/protocol.c +2 -1
@@ -323,9 +323,10 @@ static bool mptcp_rmem_schedule(struct s
 	struct mptcp_sock *msk = mptcp_sk(sk);
 	int amt, amount;
 
-	if (size < msk->rmem_fwd_alloc)
+	if (size <= msk->rmem_fwd_alloc)
 		return true;
 
+	size -= msk->rmem_fwd_alloc;
 	amt = sk_mem_pages(size);
 	amount = amt << SK_MEM_QUANTUM_SHIFT;
 	msk->rmem_fwd_alloc += amount;
author maintainer
#1085 lore raw

PATCH 1083/1095 tracing: Use a copy of the va_list for __assign_vstr()

In reply to Greg Kroah-Hartman

From: Steven Rostedt (Google) <[email protected]>

commit 3a2dcbaf4d31023106975d6ae75b6df080c454cb upstream.

If an instance of tracing enables the same trace event as another instance, or the top level instance, or even perf, then the va_list passed into some tracepoints can be used more than once.

As va_list can only be traversed once, this can cause issues:

 # cat /sys/kernel/tracing/instances/qla2xxx/trace
             cat-56106   [012] ..... 2419873.470098: ql_dbg_log: qla2xxx [0000:05:00.0]-1054:14:  Entered (null).
             cat-56106   [012] ..... 2419873.470101: ql_dbg_log: qla2xxx [0000:05:00.0]-1000:14:  Entered ×+<96>²Ü<98>^H.
             cat-56106   [012] ..... 2419873.470102: ql_dbg_log: qla2xxx [0000:05:00.0]-1006:14:  Prepare to issue mbox cmd=0xde589000.
 # cat /sys/kernel/tracing/trace
             cat-56106   [012] ..... 2419873.470097: ql_dbg_log: qla2xxx [0000:05:00.0]-1054:14:  Entered qla2x00_get_firmware_state.
             cat-56106   [012] ..... 2419873.470100: ql_dbg_log: qla2xxx [0000:05:00.0]-1000:14:  Entered qla2x00_mailbox_command.
             cat-56106   [012] ..... 2419873.470102: ql_dbg_log: qla2xxx [0000:05:00.0]-1006:14:  Prepare to issue mbox cmd=0x69.

The instance version is corrupted because the top level instance iterated the va_list first.

Use va_copy() in the __assign_vstr() macro to make sure that each trace event for each use case gets a fresh va_list.

Fixes 0563231f93c6d ("tracing/events: Add __vstring() and __assign_vstr() helper macros")
---
 include/trace/stages/stage6_event_callback.h | 7 ++++++-
 1 file changed, 6 insertions(+), 1 deletion(-)
include/trace/stages/stage6_event_callback.h +6 -1
index 0f51f6b3ab70..3c554a585320 100644
@@ -40,7 +40,12 @@
 
 #undef __assign_vstr
 #define __assign_vstr(dst, fmt, va)					\
-	vsnprintf(__get_str(dst), TRACE_EVENT_STR_MAX, fmt, *(va))
+	do {								\
+		va_list __cp_va;					\
+		va_copy(__cp_va, *(va));				\
+		vsnprintf(__get_str(dst), TRACE_EVENT_STR_MAX, fmt, __cp_va); \
+		va_end(__cp_va);					\
+	} while (0)
 
 #undef __bitmask
 #define __bitmask(item, nr_bits) __dynamic_array(unsigned long, item, -1)
Signature · 1 line
2.37.2
author maintainer
#1086 lore raw

PATCH 1084/1095 net: dsa: felix: fix min gate len calculation for tc when its first gate is closed

In reply to Greg Kroah-Hartman

From: Vladimir Oltean <[email protected]>

commit 7e4babffa6f340a74c820d44d44d16511e666424 upstream.

min_gate_len[tc] is supposed to track the shortest interval of continuously open gates for a traffic class. For example, in the following case:

TC 76543210

t0 00000001b 200000 ns
t1 00000010b 200000 ns

min_gate_len[0] and min_gate_len[1] should be 200000, while min_gate_len[2-7] should be 0.

However what happens is that min_gate_len[0] is 200000, but min_gate_len[1] ends up being 0 (despite gate_len[1] being 200000 at the point where the logic detects the gate close event for TC 1).

The problem is that the code considers a "gate close" event whenever it sees that there is a 0 for that TC (essentially it's level rather than edge triggered). By doing that, any time a gate is seen as closed without having been open prior, gate_len, which is 0, will be written into min_gate_len. Once min_gate_len becomes 0, it's impossible for it to track anything higher than that (the length of actually open intervals).

To fix this, we make the writing to min_gate_len[tc] be edge-triggered, which avoids writes for gates that are closed in consecutive intervals. However what this does is it makes us need to special-case the permanently closed gates at the end.

Fixes 55a515b1f5a9 ("net: dsa: felix: drop oversized frames with tc-taprio instead of hanging the port")
---
 drivers/net/dsa/ocelot/felix_vsc9959.c |   15 ++++++++++++++-
 1 file changed, 14 insertions(+), 1 deletion(-)
drivers/net/dsa/ocelot/felix_vsc9959.c +14 -1
@@ -1133,6 +1133,7 @@ static void vsc9959_tas_min_gate_lengths
 {
 	struct tc_taprio_sched_entry *entry;
 	u64 gate_len[OCELOT_NUM_TC];
+	u8 gates_ever_opened = 0;
 	int tc, i, n;
 
 	/* Initialize arrays */
@@ -1160,16 +1161,28 @@ static void vsc9959_tas_min_gate_lengths
 		for (tc = 0; tc < OCELOT_NUM_TC; tc++) {
 			if (entry->gate_mask & BIT(tc)) {
 				gate_len[tc] += entry->interval;
+				gates_ever_opened |= BIT(tc);
 			} else {
 				/* Gate closes now, record a potential new
 				 * minimum and reinitialize length
 				 */
-				if (min_gate_len[tc] > gate_len[tc])
+				if (min_gate_len[tc] > gate_len[tc] &&
+				    gate_len[tc])
 					min_gate_len[tc] = gate_len[tc];
 				gate_len[tc] = 0;
 			}
 		}
 	}
+
+	/* min_gate_len[tc] actually tracks minimum *open* gate time, so for
+	 * permanently closed gates, min_gate_len[tc] will still be U64_MAX.
+	 * Therefore they are currently indistinguishable from permanently
+	 * open gates. Overwrite the gate len with 0 when we know they're
+	 * actually permanently closed, i.e. after the loop above.
+	 */
+	for (tc = 0; tc < OCELOT_NUM_TC; tc++)
+		if (!(gates_ever_opened & BIT(tc)))
+			min_gate_len[tc] = 0;
 }
 
 /* Update QSYS_PORT_MAX_SDU to make sure the static guard bands added by the
author maintainer
#1087 lore raw

PATCH 1085/1095 Revert "s390/smp: enforce lowcore protection on CPU restart"

In reply to Greg Kroah-Hartman

From: Alexander Gordeev <[email protected]>

commit 953503751a426413ea8aee2299ae3ee971b70d9b upstream.

This reverts commit 6f5c672d17f583b081e283927f5040f726c54598.

This breaks normal crash dump when CPU0 is offline.

---
 arch/s390/kernel/setup.c |    2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)
arch/s390/kernel/setup.c +1 -1
@@ -508,8 +508,8 @@ static void __init setup_lowcore_dat_on(
 	S390_lowcore.svc_new_psw.mask |= PSW_MASK_DAT;
 	S390_lowcore.program_new_psw.mask |= PSW_MASK_DAT;
 	S390_lowcore.io_new_psw.mask |= PSW_MASK_DAT;
-	__ctl_set_bit(0, 28);
 	__ctl_store(S390_lowcore.cregs_save_area, 0, 15);
+	__ctl_set_bit(0, 28);
 	put_abs_lowcore(restart_flags, RESTART_FLAG_CTLREGS);
 	put_abs_lowcore(program_new_psw, lc->program_new_psw);
 	for (cr = 0; cr < ARRAY_SIZE(lc->cregs_save_area); cr++)
author maintainer
#1088 lore raw

PATCH 1086/1095 powerpc/kexec: Fix build failure from uninitialised variable

In reply to Greg Kroah-Hartman

From: Russell Currey <[email protected]>

commit 83ee9f23763a432a4077bf20624ee35de87bce99 upstream.

clang 14 won't build because ret is uninitialised and can be returned if both prop and fdtprop are NULL. Drop the ret variable and return an error in that failure case.

Fixes b1fc44eaa9ba ("pseries/iommu/ddw: Fix kdump to work in absence of ibm,dma-window")
---
 arch/powerpc/kexec/file_load_64.c |   10 +++++-----
 1 file changed, 5 insertions(+), 5 deletions(-)
arch/powerpc/kexec/file_load_64.c +5 -5
@@ -1043,17 +1043,17 @@ static int copy_property(void *fdt, int
 			 const char *propname)
 {
 	const void *prop, *fdtprop;
-	int len = 0, fdtlen = 0, ret;
+	int len = 0, fdtlen = 0;
 
 	prop = of_get_property(dn, propname, &len);
 	fdtprop = fdt_getprop(fdt, node_offset, propname, &fdtlen);
 
 	if (fdtprop && !prop)
-		ret = fdt_delprop(fdt, node_offset, propname);
+		return fdt_delprop(fdt, node_offset, propname);
 	else if (prop)
-		ret = fdt_setprop(fdt, node_offset, propname, prop, len);
-
-	return ret;
+		return fdt_setprop(fdt, node_offset, propname, prop, len);
+	else
+		return -FDT_ERR_NOTFOUND;
 }
 
 static int update_pci_dma_nodes(void *fdt, const char *dmapropname)
author maintainer
#1089 lore raw

PATCH 1087/1095 drm/bridge: tc358767: Fix (e)DP bridge endpoint parsing in dedicated function

In reply to Greg Kroah-Hartman

From: Marek Vasut <[email protected]>

commit 9030a9e571b3ba250d3d450a98310e3c74ecaff4 upstream.

Per toshiba,tc358767.yaml DT binding document, port@2 the output (e)DP port is optional. In case this port is not described in DT, the bridge driver operates in DPI-to-DP mode. The drm_of_find_panel_or_bridge() call in tc_probe_edp_bridge_endpoint() returns -ENODEV in case port@2 is not present in DT and this specific return value is incorrectly propagated outside of tc_probe_edp_bridge_endpoint() function. All other error values must be propagated and are propagated correctly.

Return 0 in case the port@2 is missing instead, that reinstates the original behavior before the commit this patch fixes.

Fixes 8478095a8c4b ("drm/bridge: tc358767: Move (e)DP bridge endpoint parsing into dedicated function")
Cc Jonas Karlman <[email protected]>
Cc Laurent Pinchart <[email protected]>
Cc Lucas Stach <[email protected]>
Cc Marek Vasut <[email protected]>
Cc Maxime Ripard <[email protected]>
Cc Neil Armstrong <[email protected]>
Cc Robert Foss <[email protected]>
Cc Sam Ravnborg <[email protected]>
---
 drivers/gpu/drm/bridge/tc358767.c |    2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)
drivers/gpu/drm/bridge/tc358767.c +1 -1
@@ -1573,7 +1573,7 @@ static int tc_probe_edp_bridge_endpoint(
 		tc->bridge.type = DRM_MODE_CONNECTOR_DisplayPort;
 	}
 
-	return ret;
+	return 0;
 }
 
 static void tc_clk_disable(void *data)
author maintainer
#1090 lore raw

PATCH 1088/1095 bpf: Fix sparse warning for bpf_kptr_xchg_proto

In reply to Greg Kroah-Hartman

From: Kumar Kartikeya Dwivedi <[email protected]>

commit 5b74c690e1c55953ec99fd9dab74f72dbee4fe95 upstream.

Kernel Test Robot complained about missing static storage class annotation for bpf_kptr_xchg_proto variable.

sparse: symbol 'bpf_kptr_xchg_proto' was not declared. Should it be static?

This caused by missing extern definition in the header. Add it to suppress the sparse warning.

---
 include/linux/bpf.h |    1 +
 1 file changed, 1 insertion(+)
include/linux/bpf.h +1 -0
@@ -2228,6 +2228,7 @@ extern const struct bpf_func_proto bpf_f
 extern const struct bpf_func_proto bpf_loop_proto;
 extern const struct bpf_func_proto bpf_strncmp_proto;
 extern const struct bpf_func_proto bpf_copy_from_user_task_proto;
+extern const struct bpf_func_proto bpf_kptr_xchg_proto;
 
 const struct bpf_func_proto *tracing_prog_func_proto(
   enum bpf_func_id func_id, const struct bpf_prog *prog);
author maintainer
#1091 lore raw

PATCH 1089/1095 bpf: Suppress passing zero to PTR_ERR warning

In reply to Greg Kroah-Hartman

From: Kumar Kartikeya Dwivedi <[email protected]>

commit 1ec5ee8c8a5a65ea377f8bea64bf4d5b743f6f79 upstream.

Kernel Test Robot complains about passing zero to PTR_ERR for the said line, suppress it by using PTR_ERR_OR_ZERO.

---
 kernel/bpf/verifier.c |    2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)
kernel/bpf/verifier.c +1 -1
@@ -5327,7 +5327,7 @@ static int process_kptr_func(struct bpf_
 		return -EINVAL;
 	}
 	if (!map_value_has_kptrs(map_ptr)) {
-		ret = PTR_ERR(map_ptr->kptr_off_tab);
+		ret = PTR_ERR_OR_ZERO(map_ptr->kptr_off_tab);
 		if (ret == -E2BIG)
 			verbose(env, "map '%s' has more than %d kptr\n", map_ptr->name,
 				BPF_MAP_VALUE_OFF_MAX);
author maintainer
#1092 lore raw

PATCH 1090/1095 net: phy: smsc: Disable Energy Detect Power-Down in interrupt mode

In reply to Greg Kroah-Hartman

From: Lukas Wunner <[email protected]>

commit 2642cc6c3bbe0900ba15bab078fd15ad8baccbc5 upstream.

Simon reports that if two LAN9514 USB adapters are directly connected without an intermediate switch, the link fails to come up and link LEDs remain dark. The issue was introduced by commit 1ce8b37241ed ("usbnet:
smsc95xx: Forward PHY interrupts to PHY driver to avoid polling").

The PHY suffers from a known erratum wherein link detection becomes unreliable if Energy Detect Power-Down is used. In poll mode, the driver works around the erratum by briefly disabling EDPD for 640 msec to detect a neighbor, then re-enabling it to save power.

In interrupt mode, no interrupt is signaled if EDPD is used by both link partners, so it must not be enabled at all.

We'll recoup the power savings by enabling SUSPEND1 mode on affected LAN95xx chips in a forthcoming commit.

---
 drivers/net/phy/smsc.c |    6 ++++--
 1 file changed, 4 insertions(+), 2 deletions(-)
drivers/net/phy/smsc.c +4 -2
@@ -110,7 +110,7 @@ static int smsc_phy_config_init(struct p
 	struct smsc_phy_priv *priv = phydev->priv;
 	int rc;
 
-	if (!priv->energy_enable)
+	if (!priv->energy_enable || phydev->irq != PHY_POLL)
 		return 0;
 
 	rc = phy_read(phydev, MII_LAN83C185_CTRL_STATUS);
@@ -210,6 +210,8 @@ static int lan95xx_config_aneg_ext(struc
  * response on link pulses to detect presence of plugged Ethernet cable.
  * The Energy Detect Power-Down mode is enabled again in the end of procedure to
  * save approximately 220 mW of power if cable is unplugged.
+ * The workaround is only applicable to poll mode. Energy Detect Power-Down may
+ * not be used in interrupt mode lest link change detection becomes unreliable.
  */
 static int lan87xx_read_status(struct phy_device *phydev)
 {
@@ -217,7 +219,7 @@ static int lan87xx_read_status(struct ph
 
 	int err = genphy_read_status(phydev);
 
-	if (!phydev->link && priv->energy_enable) {
+	if (!phydev->link && priv->energy_enable && phydev->irq == PHY_POLL) {
 		/* Disable EDPD to wake up PHY */
 		int rc = phy_read(phydev, MII_LAN83C185_CTRL_STATUS);
 		if (rc < 0)
author maintainer
#1093 lore raw

PATCH 1091/1095 f2fs: revive F2FS_IOC_ABORT_VOLATILE_WRITE

In reply to Greg Kroah-Hartman

From: Daeho Jeong <[email protected]>

commit 23339e5752d01a4b5e122759b002cf896d26f6c1 upstream.

F2FS_IOC_ABORT_VOLATILE_WRITE was used to abort a atomic write before. However it was removed accidentally. So revive it by changing the name, since volatile write had gone.

Fiexes: 7bc155fec5b3("f2fs: kill volatile write support")

---
 fs/f2fs/file.c            |   30 ++++++++++++++++++++++++++++--
 include/uapi/linux/f2fs.h |    2 +-
 2 files changed, 29 insertions(+), 3 deletions(-)
fs/f2fs/file.c +28 -2
@@ -2102,6 +2102,31 @@ unlock_out:
 	return ret;
 }
 
+static int f2fs_ioc_abort_atomic_write(struct file *filp)
+{
+	struct inode *inode = file_inode(filp);
+	struct user_namespace *mnt_userns = file_mnt_user_ns(filp);
+	int ret;
+
+	if (!inode_owner_or_capable(mnt_userns, inode))
+		return -EACCES;
+
+	ret = mnt_want_write_file(filp);
+	if (ret)
+		return ret;
+
+	inode_lock(inode);
+
+	if (f2fs_is_atomic_file(inode))
+		f2fs_abort_atomic_write(inode, true);
+
+	inode_unlock(inode);
+
+	mnt_drop_write_file(filp);
+	f2fs_update_time(F2FS_I_SB(inode), REQ_TIME);
+	return ret;
+}
+
 static int f2fs_ioc_shutdown(struct file *filp, unsigned long arg)
 {
 	struct inode *inode = file_inode(filp);
@@ -4039,9 +4064,10 @@ static long __f2fs_ioctl(struct file *fi
 		return f2fs_ioc_start_atomic_write(filp);
 	case F2FS_IOC_COMMIT_ATOMIC_WRITE:
 		return f2fs_ioc_commit_atomic_write(filp);
+	case F2FS_IOC_ABORT_ATOMIC_WRITE:
+		return f2fs_ioc_abort_atomic_write(filp);
 	case F2FS_IOC_START_VOLATILE_WRITE:
 	case F2FS_IOC_RELEASE_VOLATILE_WRITE:
-	case F2FS_IOC_ABORT_VOLATILE_WRITE:
 		return -EOPNOTSUPP;
 	case F2FS_IOC_SHUTDOWN:
 		return f2fs_ioc_shutdown(filp, arg);
@@ -4666,7 +4692,7 @@ long f2fs_compat_ioctl(struct file *file
 	case F2FS_IOC_COMMIT_ATOMIC_WRITE:
 	case F2FS_IOC_START_VOLATILE_WRITE:
 	case F2FS_IOC_RELEASE_VOLATILE_WRITE:
-	case F2FS_IOC_ABORT_VOLATILE_WRITE:
+	case F2FS_IOC_ABORT_ATOMIC_WRITE:
 	case F2FS_IOC_SHUTDOWN:
 	case FITRIM:
 	case FS_IOC_SET_ENCRYPTION_POLICY:
include/uapi/linux/f2fs.h +1 -1
@@ -13,7 +13,7 @@
 #define F2FS_IOC_COMMIT_ATOMIC_WRITE	_IO(F2FS_IOCTL_MAGIC, 2)
 #define F2FS_IOC_START_VOLATILE_WRITE	_IO(F2FS_IOCTL_MAGIC, 3)
 #define F2FS_IOC_RELEASE_VOLATILE_WRITE	_IO(F2FS_IOCTL_MAGIC, 4)
-#define F2FS_IOC_ABORT_VOLATILE_WRITE	_IO(F2FS_IOCTL_MAGIC, 5)
+#define F2FS_IOC_ABORT_ATOMIC_WRITE	_IO(F2FS_IOCTL_MAGIC, 5)
 #define F2FS_IOC_GARBAGE_COLLECT	_IOW(F2FS_IOCTL_MAGIC, 6, __u32)
 #define F2FS_IOC_WRITE_CHECKPOINT	_IO(F2FS_IOCTL_MAGIC, 7)
 #define F2FS_IOC_DEFRAGMENT		_IOWR(F2FS_IOCTL_MAGIC, 8,	\
author maintainer
#1094 lore raw

PATCH 1092/1095 drm/vc4: change vc4_dma_range_matches from a global to static

In reply to Greg Kroah-Hartman

From: Tom Rix <[email protected]>

commit 63569d90863ff26c8b10c8971d1271c17a45224b upstream.

sparse reports
drivers/gpu/drm/vc4/vc4_drv.c:270:27: warning: symbol 'vc4_dma_range_matches' was not declared. Should it be static?

vc4_dma_range_matches is only used in vc4_drv.c, so it's storage class specifier should be static.

Fixes da8e393e23ef ("drm/vc4: drv: Adopt the dma configuration from the HVS or V3D component")
---
 drivers/gpu/drm/vc4/vc4_drv.c |    2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)
drivers/gpu/drm/vc4/vc4_drv.c +1 -1
@@ -209,7 +209,7 @@ static void vc4_match_add_drivers(struct
 	}
 }
 
-const struct of_device_id vc4_dma_range_matches[] = {
+static const struct of_device_id vc4_dma_range_matches[] = {
 	{ .compatible = "brcm,bcm2711-hvs" },
 	{ .compatible = "brcm,bcm2835-hvs" },
 	{ .compatible = "brcm,bcm2835-v3d" },
author maintainer
#1095 lore raw

PATCH 1093/1095 f2fs: fix null-ptr-deref in f2fs_get_dnode_of_data

In reply to Greg Kroah-Hartman

From: Ye Bin <[email protected]>

commit 4a2c5b7994960fac29cf8a3f4e62855bae1b27d4 upstream.

There is issue as follows when test f2fs atomic write:
F2FS-fs (loop0): Can't find valid F2FS filesystem in 2th superblock F2FS-fs (loop0): invalid crc_offset: 0
F2FS-fs (loop0): f2fs_check_nid_range: out-of-range nid=1, run fsck to fix. F2FS-fs (loop0): f2fs_check_nid_range: out-of-range nid=2, run fsck to fix.
==================================================================

BUG: KASAN: null-ptr-deref in f2fs_get_dnode_of_data+0xac/0x16d0
Read of size 8 at addr 0000000000000028 by task rep/1990

CPU: 4 PID: 1990 Comm: rep Not tainted 5.19.0-rc6-next-20220715 #266
Call Trace:
 <TASK>
 dump_stack_lvl+0x6e/0x91
 print_report.cold+0x49a/0x6bb
 kasan_report+0xa8/0x130
 f2fs_get_dnode_of_data+0xac/0x16d0
 f2fs_do_write_data_page+0x2a5/0x1030
 move_data_page+0x3c5/0xdf0
 do_garbage_collect+0x2015/0x36c0
 f2fs_gc+0x554/0x1d30
 f2fs_balance_fs+0x7f5/0xda0
 f2fs_write_single_data_page+0xb66/0xdc0
 f2fs_write_cache_pages+0x716/0x1420
 f2fs_write_data_pages+0x84f/0x9a0
 do_writepages+0x130/0x3a0
 filemap_fdatawrite_wbc+0x87/0xa0
 file_write_and_wait_range+0x157/0x1c0
 f2fs_do_sync_file+0x206/0x12d0
 f2fs_sync_file+0x99/0xc0
 vfs_fsync_range+0x75/0x140
 f2fs_file_write_iter+0xd7b/0x1850
 vfs_write+0x645/0x780
 ksys_write+0xf1/0x1e0
 do_syscall_64+0x3b/0x90
 entry_SYSCALL_64_after_hwframe+0x63/0xcd

As 3db1de0e582c commit changed atomic write way which new a cow_inode for atomic write file, and also mark cow_inode as FI_ATOMIC_FILE. When f2fs_do_write_data_page write cow_inode will use cow_inode's cow_inode which is NULL. Then will trigger null-ptr-deref.
To solve above issue, introduce FI_COW_FILE flag for COW inode.

Fiexes: 3db1de0e582c("f2fs: change the current atomic write way")

---
 fs/f2fs/f2fs.h    |    6 ++++++
 fs/f2fs/file.c    |    2 +-
 fs/f2fs/segment.c |    4 ++--
 3 files changed, 9 insertions(+), 3 deletions(-)
fs/f2fs/f2fs.h +6 -0
@@ -756,6 +756,7 @@ enum {
 	FI_ENABLE_COMPRESS,	/* enable compression in "user" compression mode */
 	FI_COMPRESS_RELEASED,	/* compressed blocks were released */
 	FI_ALIGNED_WRITE,	/* enable aligned write */
+	FI_COW_FILE,		/* indicate COW file */
 	FI_MAX,			/* max flag, never be used */
 };
 
@@ -3188,6 +3189,11 @@ static inline bool f2fs_is_atomic_file(s
 	return is_inode_flag_set(inode, FI_ATOMIC_FILE);
 }
 
+static inline bool f2fs_is_cow_file(struct inode *inode)
+{
+	return is_inode_flag_set(inode, FI_COW_FILE);
+}
+
 static inline bool f2fs_is_first_block_written(struct inode *inode)
 {
 	return is_inode_flag_set(inode, FI_FIRST_BLOCK_WRITTEN);
fs/f2fs/file.c +1 -1
@@ -2055,7 +2055,7 @@ static int f2fs_ioc_start_atomic_write(s
 	spin_unlock(&sbi->inode_lock[ATOMIC_FILE]);
 
 	set_inode_flag(inode, FI_ATOMIC_FILE);
-	set_inode_flag(fi->cow_inode, FI_ATOMIC_FILE);
+	set_inode_flag(fi->cow_inode, FI_COW_FILE);
 	clear_inode_flag(fi->cow_inode, FI_INLINE_DATA);
 	f2fs_up_write(&F2FS_I(inode)->i_gc_rwsem[WRITE]);
 
fs/f2fs/segment.c +2 -2
@@ -193,7 +193,7 @@ void f2fs_abort_atomic_write(struct inod
 	if (f2fs_is_atomic_file(inode)) {
 		if (clean)
 			truncate_inode_pages_final(inode->i_mapping);
-		clear_inode_flag(fi->cow_inode, FI_ATOMIC_FILE);
+		clear_inode_flag(fi->cow_inode, FI_COW_FILE);
 		iput(fi->cow_inode);
 		fi->cow_inode = NULL;
 		clear_inode_flag(inode, FI_ATOMIC_FILE);
@@ -3167,7 +3167,7 @@ static int __get_segment_type_6(struct f
 			return CURSEG_COLD_DATA;
 		if (file_is_hot(inode) ||
 				is_inode_flag_set(inode, FI_HOT_DATA) ||
-				f2fs_is_atomic_file(inode))
+				f2fs_is_cow_file(inode))
 			return CURSEG_HOT_DATA;
 		return f2fs_rw_hint_to_seg_type(inode->i_write_hint);
 	} else {
author maintainer
#1096 lore raw

PATCH 1094/1095 io_uring: mem-account pbuf buckets

In reply to Greg Kroah-Hartman

From: Pavel Begunkov <[email protected]>

commit cc18cc5e82033d406f54144ad6f8092206004684 upstream.

Potentially, someone may create as many pbuf bucket as there are indexes in an xarray without any other restrictions bounding our memory usage, put memory needed for the buckets under memory accounting.

---
 io_uring/io_uring.c |    2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)
io_uring/io_uring.c +1 -1
@@ -4927,7 +4927,7 @@ static int io_provide_buffers(struct io_
 
 	bl = io_buffer_get_list(ctx, p->bgid);
 	if (unlikely(!bl)) {
-		bl = kmalloc(sizeof(*bl), GFP_KERNEL);
+		bl = kzalloc(sizeof(*bl), GFP_KERNEL_ACCOUNT);
 		if (!bl) {
 			ret = -ENOMEM;
 			goto err;
author maintainer
#1097 lore raw

PATCH 1095/1095 Bluetooth: L2CAP: Fix l2cap_global_chan_by_psm regression

In reply to Greg Kroah-Hartman

From: Luiz Augusto von Dentz <[email protected]>

commit 332f1795ca202489c665a75e62e18ff6284de077 upstream.

The patch d0be8347c623: "Bluetooth: L2CAP: Fix use-after-free caused by l2cap_chan_put" from Jul 21, 2022, leads to the following Smatch static checker warning:

net/bluetooth/l2cap_core.c:1977 l2cap_global_chan_by_psm()
error: we previously assumed 'c' could be null (see line 1996)

Fixes d0be8347c623 ("Bluetooth: L2CAP: Fix use-after-free caused by l2cap_chan_put")
---
 net/bluetooth/l2cap_core.c |   13 ++++++-------
 1 file changed, 6 insertions(+), 7 deletions(-)
net/bluetooth/l2cap_core.c +6 -7
@@ -1969,11 +1969,11 @@ static struct l2cap_chan *l2cap_global_c
 						   bdaddr_t *dst,
 						   u8 link_type)
 {
-	struct l2cap_chan *c, *c1 = NULL;
+	struct l2cap_chan *c, *tmp, *c1 = NULL;
 
 	read_lock(&chan_list_lock);
 
-	list_for_each_entry(c, &chan_list, global_l) {
+	list_for_each_entry_safe(c, tmp, &chan_list, global_l) {
 		if (state && c->state != state)
 			continue;
 
@@ -1992,11 +1992,10 @@ static struct l2cap_chan *l2cap_global_c
 			dst_match = !bacmp(&c->dst, dst);
 			if (src_match && dst_match) {
 				c = l2cap_chan_hold_unless_zero(c);
-				if (!c)
-					continue;
-
-				read_unlock(&chan_list_lock);
-				return c;
+				if (c) {
+					read_unlock(&chan_list_lock);
+					return c;
+				}
 			}
 
 			/* Closest match */
Tested-by on the v1 series
#1098 lore raw
In reply to Greg Kroah-Hartman
Quoted 9 lines · Greg Kroah-Hartman

This is the start of the stable review cycle for the 5.18.18 release. There are 1095 patches in this series, all will be posted as a response to this one. If anyone has any issues with these being applied, please let me know.

Responses should be made by Wed, 17 Aug 2022 18:01:29 +0000. Anything received after that time might be too late.

The whole patch series can be found in one patch at:

Quoted 9 lines
	https://www.kernel.org/pub/linux/kernel/v5.x/stable-review/patch-5.18.18-rc1.gz

or in the git tree and branch at:
git://git.kernel.org/pub/scm/linux/kernel/git/stable/linux-stable-rc.git linux-5.18.y
and the diffstat can be found below.

thanks,

greg k-h

Compiled and booted on my test system. No dmesg regressions.

thanks,
-- Shuah

Tested-by Linux Kernel Functional Testing on the v1 series
#1099 lore raw
In reply to Greg Kroah-Hartman
Quoted 10 lines · Greg Kroah-Hartman

This is the start of the stable review cycle for the 5.18.18 release. There are 1095 patches in this series, all will be posted as a response to this one. If anyone has any issues with these being applied, please let me know.

Responses should be made by Wed, 17 Aug 2022 18:01:29 +0000. Anything received after that time might be too late.

The whole patch series can be found in one patch at:

Quoted 8 lines
        https://www.kernel.org/pub/linux/kernel/v5.x/stable-review/patch-5.18.18-rc1.gz

or in the git tree and branch at:
git://git.kernel.org/pub/scm/linux/kernel/git/stable/linux-stable-rc.git linux-5.18.y
and the diffstat can be found below.

thanks,

greg k-h

Results from Linaro's test farm.
No regressions on arm64, arm, x86_64, and i386.

## Build
* kernel: 5.18.18-rc1
* git: https://gitlab.com/Linaro/lkft/mirrors/stable/linux-stable-rc
* git branch: linux-5.18.y
* git commit: 56bd9e4c28e9ed65e07e7148e65adfdf1f1c5524
* git describe: v5.18.17-1096-g56bd9e4c28e9
* test details:
https://qa-reports.linaro.org/lkft/linux-stable-rc-linux-5.18.y/build/v5.18.17-1096-g56bd9e4c28e9

## No test Regressions (compared to v5.18.16-36-g732bf05a92ab)

## No metric Regressions (compared to v5.18.16-36-g732bf05a92ab)

## No test Fixes (compared to v5.18.16-36-g732bf05a92ab)

## No metric Fixes (compared to v5.18.16-36-g732bf05a92ab)

## Test result summary
total: 93607, pass: 85358, fail: 732, skip: 6930, xfail: 587

## Build Summary
* arc: 10 total, 10 passed, 0 failed
* arm: 301 total, 301 passed, 0 failed
* arm64: 62 total, 60 passed, 2 failed
* i386: 52 total, 50 passed, 2 failed
* mips: 45 total, 45 passed, 0 failed
* parisc: 12 total, 12 passed, 0 failed
* powerpc: 60 total, 54 passed, 6 failed
* riscv: 27 total, 22 passed, 5 failed
* s390: 18 total, 18 passed, 0 failed
* sh: 24 total, 24 passed, 0 failed
* sparc: 12 total, 12 passed, 0 failed
* x86_64: 55 total, 53 passed, 2 failed

## Test suites summary
* fwts
* kunit
* kvm-unit-tests
* libgpiod
* libhugetlbfs
* log-parser-boot
* log-parser-test
* ltp-cap_bounds
* ltp-commands
* ltp-containers
* ltp-controllers
* ltp-cpuhotplug
* ltp-crypto
* ltp-cve
* ltp-dio
* ltp-fcntl-locktests
* ltp-filecaps
* ltp-fs
* ltp-fs_bind
* ltp-fs_perms_simple
* ltp-fsx
* ltp-hugetlb
* ltp-io
* ltp-ipc
* ltp-math
* ltp-mm
* ltp-nptl
* ltp-open-posix-tests
* ltp-pty
* ltp-sched
* ltp-securebits
* ltp-smoke
* ltp-syscalls
* ltp-tracing
* network-basic-tests
* packetdrill
* rcutorture
* ssuite
* v4l2-compliance
* vdso

--
Linaro LKFT
https://lkft.linaro.org

#1100 lore raw
In reply to Greg Kroah-Hartman

Hi Greg,

Quoted 8 lines · Greg Kroah-Hartman

This is the start of the stable review cycle for the 5.18.18 release. There are 1095 patches in this series, all will be posted as a response to this one. If anyone has any issues with these being applied, please let me know.

Responses should be made by Wed, 17 Aug 2022 18:01:29 +0000. Anything received after that time might be too late.

Both v5.18.18-rc1 and v5.15.61-rc1 has hung task in RPI4B causing a huge delay in bootup.

This is from v5.18.18-rc1.

[  846.815074] INFO: task systemd-udevd:192 blocked for more than 724 seconds.
[  846.822452]       Tainted: G        WC        5.18.18-rc1 #1
[  846.828961] "echo 0 > /proc/sys/kernel/hung_task_timeout_secs"
disables this message.
[  846.837557] task:systemd-udevd   state:D stack:    0 pid:  192
ppid:   177 flags:0x00000009
[  846.837582] Call trace:
[  846.837589]  __switch_to+0xf8/0x150
[  846.837611]  __schedule+0x3cc/0xa0c
[  846.837623]  schedule+0x68/0x104
[  846.837634]  schedule_preempt_disabled+0x30/0x4c
[  846.837646]  __mutex_lock.constprop.0+0x168/0x55c
[  846.837658]  __mutex_lock_slowpath+0x1c/0x30
[  846.837669]  mutex_lock+0x50/0x60
[  846.837680]  usb_udc_uevent+0x58/0xf0 [udc_core]
[  846.837718]  dev_uevent+0x134/0x2e0
[  846.837732]  kobject_uevent_env+0x298/0x764
[  846.837744]  kobject_uevent+0x14/0x20
[  846.837753]  device_add+0x398/0x8a0
[  846.837764]  usb_add_gadget+0x110/0x1bc [udc_core]
[  846.837791]  usb_add_gadget_udc+0x7c/0xc0 [udc_core]
[  846.837817]  dwc2_driver_probe+0x5e0/0x7a0 [dwc2]
[  846.837869]  platform_probe+0x74/0xd0
[  846.837880]  really_probe+0x180/0x3cc
[  846.837893]  __driver_probe_device+0x11c/0x190
[  846.837905]  driver_probe_device+0x44/0xf4
[  846.837917]  __driver_attach+0xd8/0x1b0
[  846.837930]  bus_for_each_dev+0x7c/0xe0
[  846.837941]  driver_attach+0x30/0x40
[  846.837953]  bus_add_driver+0x154/0x240
[  846.837964]  driver_register+0x84/0x140
[  846.837977]  __platform_driver_register+0x34/0x40
[  846.837987]  dwc2_platform_driver_init+0x2c/0x1000 [dwc2]
[  846.838035]  do_one_initcall+0x50/0x2c0
[  846.838047]  do_init_module+0x50/0x260
[  846.838061]  load_module+0x2298/0x2720
[  846.838074]  __do_sys_finit_module+0xac/0x12c
[  846.838087]  __arm64_sys_finit_module+0x2c/0x40
[  846.838099]  invoke_syscall+0x50/0x120
[  846.838113]  el0_svc_common.constprop.0+0x6c/0x1a0
[  846.838126]  do_el0_svc+0x30/0x90
[  846.838138]  el0_svc+0x30/0xd0
[  846.838147]  el0t_64_sync_handler+0x10c/0x140
[  846.838158]  el0t_64_sync+0x1a0/0x1a4
[  846.838179] INFO: task udevadm:964 blocked for more than 362 seconds.
[  846.845473]       Tainted: G        WC        5.18.18-rc1 #1
[  846.851891] "echo 0 > /proc/sys/kernel/hung_task_timeout_secs"
disables this message.
[  846.860543] task:udevadm         state:D stack:    0 pid:  964
ppid:     1 flags:0x00000000
[  846.860555] Call trace:
[  846.860560]  __switch_to+0xf8/0x150
[  846.860569]  __schedule+0x3cc/0xa0c
[  846.860576]  schedule+0x68/0x104
[  846.860582]  schedule_preempt_disabled+0x30/0x4c
[  846.860589]  __mutex_lock.constprop.0+0x168/0x55c
[  846.860596]  __mutex_lock_slowpath+0x1c/0x30
[  846.860603]  mutex_lock+0x50/0x60
[  846.860610]  usb_udc_uevent+0x58/0xf0 [udc_core]
[  846.860629]  dev_uevent+0x134/0x2e0
[  846.860637]  uevent_show+0x98/0x124
[  846.860643]  dev_attr_show+0x2c/0x6c
[  846.860649]  sysfs_kf_seq_show+0x98/0x110
[  846.860656]  kernfs_seq_show+0x38/0x44
[  846.860664]  seq_read_iter+0x170/0x48c
[  846.860674]  kernfs_fop_read_iter+0x198/0x1d4
[  846.860681]  new_sync_read+0xd8/0x15c
[  846.860689]  vfs_read+0x19c/0x1e4
[  846.860697]  ksys_read+0x78/0x110
[  846.860704]  __arm64_sys_read+0x28/0x34
[  846.860711]  invoke_syscall+0x50/0x120
[  846.860719]  el0_svc_common.constprop.0+0x6c/0x1a0
[  846.860727]  do_el0_svc+0x30/0x90
[  846.860734]  el0_svc+0x30/0xd0
[  846.860740]  el0t_64_sync_handler+0x10c/0x140
[  846.860746]  el0t_64_sync+0x1a0/0x1a4

Apart from the call trace from the hung task I can also see another warning.

[   14.841402] ------------[ cut here ]------------
[   14.845855] Bluetooth: HCI UART protocol ATH3K registered
[   14.851200] WARNING: CPU: 0 PID: 184 at
drivers/gpu/drm/vc4/vc4_hdmi_regs.h:450 vc5_hdmi_reset+0x1f4/0x234
[vc4]
[   14.867061] Bluetooth: HCI UART protocol Three-wire (H5) registered
[   14.876900] Modules linked in: brcmfmac(+) hci_uart(+) brcmutil

btqca bcm2835_v4l2(C) btrtl btbcm btintel bcm2835_mmal_vchiq(C) videobuf2_vmalloc vc4(+) raspberrypi_hwmon cfg80211 cec videobuf2_memops videobuf2_v4l2 drm_cma_helper videobuf2_common bluetooth xhci_pci(+) drm_kms_helper drm dwc2(+) udc_core ecdh_generic snd_soc_core snd_bcm2835(C) i2c_brcmstb roles videodev ecc ac97_bus snd_pcm_dmaengine snd_pcm pwm_bcm2835 mc snd_timer xhci_pci_renesas snd fb_sys_fops syscopyarea sysfillrect crct10dif_ce phy_generic uio_pdrv_genirq sysimgblt uio aes_neon_bs aes_neon_blk

[   14.934380] CPU: 0 PID: 184 Comm: systemd-udevd Tainted: G
C        5.18.18-rc1 #1
[   14.942761] Hardware name: Raspberry Pi 4 Model B (DT)
[   14.947969] pstate: 604000c5 (nZCv daIF +PAN -UAO -TCO -DIT -SSBS BTYPE=--)
[   14.955031] pc : vc5_hdmi_reset+0x1f4/0x234 [vc4]
[   14.959837] lr : vc5_hdmi_reset+0x38/0x234 [vc4]
[   14.964548] sp : ffff80000a7b3640
[   14.967902] x29: ffff80000a7b3640 x28: 0000000000000000 x27: ffff000056c35080
[   14.975140] x26: ffff80000136db78 x25: ffff80000136e078 x24: 0000000000000000
[   14.976233] Bluetooth: HCI UART protocol Intel registered
[   14.982376] x23: ffff000058479000 x22: ffff0000fb836da0 x21: ffff000056c35c90
[   14.982383] x20: 0000000000000000 x19: ffff000056c35080 x18: 0000000000000014
[   14.982389] x17: 00000000f19a3ea9 x16: 00000000b24d2d64 x15: 000000006490e735
[   14.982396] x14: 0000000000000000 x13: 00000000d43c985e x12: 00000000f1020f23
[   14.982402] x11: 00000000d1b738e5 x10: ffff8400035fbc7f x9 : ffff800001357768
[   14.982408] x8 : 0101010101010101 x7 : 0000000000000000 x6 : ffff000055ade180
[   14.982414] x5 : 0000000000000000 x4 : 0000000000000001 x3 : 0000000000000000
[   14.982420] x2 : 0000000000000001 x1 : 0000000000000002 x0 : ffff80000136f7f0
[   15.026130] Bluetooth: HCI UART protocol Broadcom registered
[   15.031261]
[   15.031263] Call trace:
[   15.031265]  vc5_hdmi_reset+0x1f4/0x234 [vc4]
[   15.041432] Bluetooth: HCI UART protocol QCA registered
[   15.045761]  vc4_hdmi_runtime_resume+0x74/0x2d0 [vc4]
[   15.045798]  vc4_hdmi_bind+0x244/0xa60 [vc4]
[   15.074641]  component_bind_all+0x114/0x26c
[   15.078887]  vc4_drm_bind+0x1d8/0x2c0 [vc4]
[   15.083159]  try_to_bring_up_aggregate_device+0x200/0x2d4
[   15.085085] Bluetooth: HCI UART protocol AG6XX registered
[   15.088635]  component_master_add_with_match+0xcc/0x110
[   15.088639]  vc4_platform_drm_probe+0xc8/0x100 [vc4]
[   15.104457]  platform_probe+0x74/0xd0
[   15.108169]  really_probe+0x180/0x3cc
[   15.111881]  __driver_probe_device+0x11c/0x190
[   15.114311] Bluetooth: HCI UART protocol Marvell registered
[   15.116384]  driver_probe_device+0x44/0xf4
[   15.116390]  __driver_attach+0xd8/0x1b0
[   15.116396]  bus_for_each_dev+0x7c/0xe0
[   15.116401]  driver_attach+0x30/0x40
[   15.116406]  bus_add_driver+0x154/0x240
[   15.116410]  driver_register+0x84/0x140
[   15.116416]  __platform_driver_register+0x34/0x40
[   15.116419]  vc4_drm_register+0x60/0x1000 [vc4]
[   15.154701]  do_one_initcall+0x50/0x2c0
[   15.158590]  do_init_module+0x50/0x260
[   15.162391]  load_module+0x2298/0x2720
[   15.166189]  __do_sys_finit_module+0xac/0x12c
[   15.170604]  __arm64_sys_finit_module+0x2c/0x40
[   15.175196]  invoke_syscall+0x50/0x120
[   15.178995]  el0_svc_common.constprop.0+0x6c/0x1a0
[   15.183851]  do_el0_svc+0x30/0x90
[   15.187208]  el0_svc+0x30/0xd0
[   15.190301]  el0t_64_sync_handler+0x10c/0x140
[   15.194715]  el0t_64_sync+0x1a0/0x1a4
[   15.198424] ---[ end trace 0000000000000000 ]---
[   15.203150] ------------[ cut here ]------------

I can see similar trace with hung task of systemd-udevd in v5.15.61-rc1 also, but not the other warning.

I am on holiday, might not be able to bisect to find the offending commit.

Signature · 2 lines
Regards
Sudip