Skip to contentSkip to conversation

drivers/base/node.c: fix userspace break from using bin_attributes for cpumap and cpulist

PATCH Superseded Phil Auld 2 participants 5 messages Jul 13, 2022 Driver core

v2 of 5 Superseded Latest is v5 →

lore
Share & tools
lore

Phil Auld

author
#0 lore raw

PATCH v2 1/1 drivers/base/node.c: fix userspace break from using bin_attributes for cpumap and cpulist

Using bin_attributes with a 0 size causes fstat and friends to return that 0 size. This breaks userspace code that retrieves the size before reading the file. Rather than reverting 75bd50fa841 ("drivers/base/node.c: use bin_attribute to break the size limitation of cpumap ABI") let's put in a size value at compile time. Use direct comparison and a worst-case maximum to ensure compile time constants. For cpulist the max is on the order of NR_CPUS * (ceil(log10(NR_CPUS)) + 1) which for 8192 is 40960 (8192 * 5). In order to get near that you'd need a system with every other CPU on one node or something similar. e.g. (0,2,4,... 1024,1026...). To simplify the math and support larger NR_CPUS we are using NR_CPUS * 6. We also set it to a min of PAGE_SIZE to retain the older behavior for smaller NR_CPUS. The cpumap file wants to be something like NR_CPUS/4 + NR_CPUS/32, for the ","s so for simplicity we are using NR_CPUS/2.

On an 80 cpu 4-node sytem (NR_CPUS == 8192)

before:

-r--r--r--. 1 root root 0 Jul 12 14:08 /sys/devices/system/node/node0/cpulist -r--r--r--. 1 root root 0 Jul 11 17:25 /sys/devices/system/node/node0/cpumap

after:

-r--r--r--. 1 root root 49152 Jul 13 09:26 /sys/devices/system/node/node0/cpulist -r--r--r--. 1 root root 4096 Jul 13 09:24 /sys/devices/system/node/node0/cpumap

Fixes 75bd50fa841 ("drivers/base/node.c: use bin_attribute to break the size limitation of cpumap ABI")
Cc Greg Kroah-Hartman <[email protected]>
Cc "Rafael J. Wysocki" <[email protected]>
---
 drivers/base/node.c | 4 ++--
 1 file changed, 2 insertions(+), 2 deletions(-)
drivers/base/node.c +2 -2
index 0ac6376ef7a1..3577f4eb4ac6 100644
@@ -45,7 +45,7 @@ static inline ssize_t cpumap_read(struct file *file, struct kobject *kobj,
 	return n;
 }
 
-static BIN_ATTR_RO(cpumap, 0);
+static BIN_ATTR_RO(cpumap, (((NR_CPUS>>1) > PAGE_SIZE) ? NR_CPUS >> 1 : PAGE_SIZE));
 
 static inline ssize_t cpulist_read(struct file *file, struct kobject *kobj,
 				   struct bin_attribute *attr, char *buf,
@@ -66,7 +66,7 @@ static inline ssize_t cpulist_read(struct file *file, struct kobject *kobj,
 	return n;
 }
 
-static BIN_ATTR_RO(cpulist, 0);
+static BIN_ATTR_RO(cpulist, (((NR_CPUS * 6) > PAGE_SIZE) ? NR_CPUS *6 : PAGE_SIZE));
 
 /**
  * struct node_access_nodes - Access class device to hold user visible
Signature · 1 line
2.31.1
#1 lore raw
Quoted 37 lines · Phil Auld

Using bin_attributes with a 0 size causes fstat and friends to return that 0 size. This breaks userspace code that retrieves the size before reading the file. Rather than reverting 75bd50fa841 ("drivers/base/node.c: use bin_attribute to break the size limitation of cpumap ABI") let's put in a size value at compile time. Use direct comparison and a worst-case maximum to ensure compile time constants. For cpulist the max is on the order of NR_CPUS * (ceil(log10(NR_CPUS)) + 1) which for 8192 is 40960 (8192 * 5). In order to get near that you'd need a system with every other CPU on one node or something similar. e.g. (0,2,4,... 1024,1026...). To simplify the math and support larger NR_CPUS we are using NR_CPUS * 6. We also set it to a min of PAGE_SIZE to retain the older behavior for smaller NR_CPUS. The cpumap file wants to be something like NR_CPUS/4 + NR_CPUS/32, for the ","s so for simplicity we are using NR_CPUS/2.

On an 80 cpu 4-node sytem (NR_CPUS == 8192)

before:

(quoted diff) +0 -2
-r--r--r--. 1 root root 0 Jul 12 14:08 /sys/devices/system/node/node0/cpulist
-r--r--r--. 1 root root 0 Jul 11 17:25 /sys/devices/system/node/node0/cpumap

after:

(quoted diff) +0 -2
-r--r--r--. 1 root root 49152 Jul 13 09:26 /sys/devices/system/node/node0/cpulist
-r--r--r--. 1 root root  4096 Jul 13 09:24 /sys/devices/system/node/node0/cpumap
Fixes 75bd50fa841 ("drivers/base/node.c: use bin_attribute to break the size limitation of cpumap ABI")
Cc Greg Kroah-Hartman <[email protected]>
Cc "Rafael J. Wysocki" <[email protected]>
---
 drivers/base/node.c | 4 ++--
 1 file changed, 2 insertions(+), 2 deletions(-)
drivers/base/node.c +0 -0
index 0ac6376ef7a1..3577f4eb4ac6 100644

drivers/base/node.c static inline ssize_t cpumap_read(struct file *file, struct kobject *kobj, · line 48

Quoted 5 lines
drivers/base/node.c +1 -1
@@ -45,7 +45,7 @@ static inline ssize_t cpumap_read(struct file *file, struct kobject *kobj,
 	return n;
 }
 
-static BIN_ATTR_RO(cpumap, 0);
+static BIN_ATTR_RO(cpumap, (((NR_CPUS>>1) > PAGE_SIZE) ? NR_CPUS >> 1 : PAGE_SIZE));

Why not just PAGE_SIZE?

If not, a HUGE comment explaining what you are doing here is going to be required.

drivers/base/node.c static inline ssize_t cpulist_read(struct file *file, struct kobject *kobj, · line 69

Quoted 9 lines
drivers/base/node.c +1 -1
 static inline ssize_t cpulist_read(struct file *file, struct kobject *kobj,
 				   struct bin_attribute *attr, char *buf,
@@ -66,7 +66,7 @@ static inline ssize_t cpulist_read(struct file *file, struct kobject *kobj,
 	return n;
 }
 
-static BIN_ATTR_RO(cpulist, 0);
+static BIN_ATTR_RO(cpulist, (((NR_CPUS * 6) > PAGE_SIZE) ? NR_CPUS *6 : PAGE_SIZE));

"* 6" in both places here?

And again, we need a huge comment explaining this so that we can understand it in a few weeks, and in 10 years...

thanks,

greg k-h

author
#2 lore raw
Quoted 41 lines · On Wed, Jul 13, 2022 at 03:58:43PM +0200 Greg Kroah-Hartman wrote:
Quoted 40 lines · Phil Auld

Using bin_attributes with a 0 size causes fstat and friends to return that 0 size. This breaks userspace code that retrieves the size before reading the file. Rather than reverting 75bd50fa841 ("drivers/base/node.c: use bin_attribute to break the size limitation of cpumap ABI") let's put in a size value at compile time. Use direct comparison and a worst-case maximum to ensure compile time constants. For cpulist the max is on the order of NR_CPUS * (ceil(log10(NR_CPUS)) + 1) which for 8192 is 40960 (8192 * 5). In order to get near that you'd need a system with every other CPU on one node or something similar. e.g. (0,2,4,... 1024,1026...). To simplify the math and support larger NR_CPUS we are using NR_CPUS * 6. We also set it to a min of PAGE_SIZE to retain the older behavior for smaller NR_CPUS. The cpumap file wants to be something like NR_CPUS/4 + NR_CPUS/32, for the ","s so for simplicity we are using NR_CPUS/2.

On an 80 cpu 4-node sytem (NR_CPUS == 8192)

before:

(quoted diff) +0 -2
-r--r--r--. 1 root root 0 Jul 12 14:08 /sys/devices/system/node/node0/cpulist
-r--r--r--. 1 root root 0 Jul 11 17:25 /sys/devices/system/node/node0/cpumap

after:

(quoted diff) +0 -2
-r--r--r--. 1 root root 49152 Jul 13 09:26 /sys/devices/system/node/node0/cpulist
-r--r--r--. 1 root root  4096 Jul 13 09:24 /sys/devices/system/node/node0/cpumap
Fixes 75bd50fa841 ("drivers/base/node.c: use bin_attribute to break the size limitation of cpumap ABI")
Cc Greg Kroah-Hartman <[email protected]>
Cc "Rafael J. Wysocki" <[email protected]>
---
 drivers/base/node.c | 4 ++--
 1 file changed, 2 insertions(+), 2 deletions(-)
drivers/base/node.c +0 -0
index 0ac6376ef7a1..3577f4eb4ac6 100644
@@ -45,7 +45,7 @@ static inline ssize_t cpumap_read(struct file *file, struct kobject *kobj,
 	return n;
 }
 
Quoted 5 lines

drivers/base/node.c static inline ssize_t cpumap_read(struct file *file, struct kobject *kobj, · line 48

Quoted 2 lines
drivers/base/node.c +1 -1
-static BIN_ATTR_RO(cpumap, 0);
+static BIN_ATTR_RO(cpumap, (((NR_CPUS>>1) > PAGE_SIZE) ? NR_CPUS >> 1 : PAGE_SIZE));

Why not just PAGE_SIZE?

This one is likely to grow beyond page size if NR_CPUS is larger than 8192. It's about 2300 bytes at 8192 if my math is right.

Let me ask again before I write the comments... what values of NR_CPUS are you seeing? Are people using an order of magnitude above 8192? Two orders? This patch (with the 6 below) handles about 1 order more.

Quoted 2 lines

If not, a HUGE comment explaining what you are doing here is going to be required.

Sure, will do. See above :)

Quoted 8 lines
Quoted 8 lines
drivers/base/node.c +0 -0
 static inline ssize_t cpulist_read(struct file *file, struct kobject *kobj,
 				   struct bin_attribute *attr, char *buf,
@@ -66,7 +66,7 @@ static inline ssize_t cpulist_read(struct file *file, struct kobject *kobj,
 	return n;
 }
 
Quoted 5 lines

drivers/base/node.c static inline ssize_t cpulist_read(struct file *file, struct kobject *kobj, · line 69

Quoted 2 lines
drivers/base/node.c +1 -1
-static BIN_ATTR_RO(cpulist, 0);
+static BIN_ATTR_RO(cpulist, (((NR_CPUS * 6) > PAGE_SIZE) ? NR_CPUS *6 : PAGE_SIZE));

"* 6" in both places here?

It's compile time so it's cheap. But yes, 6 in both places. That's to support NR_CPUS=100000.

Quoted 2 lines

And again, we need a huge comment explaining this so that we can understand it in a few weeks, and in 10 years...

Will do.

Thanks,
Phil

Quoted 5 lines

thanks,

greg k-h

#3 lore raw
Quoted 48 lines · Phil Auld
Quoted 41 lines · On Wed, Jul 13, 2022 at 03:58:43PM +0200 Greg Kroah-Hartman wrote:
Quoted 40 lines · Phil Auld

Using bin_attributes with a 0 size causes fstat and friends to return that 0 size. This breaks userspace code that retrieves the size before reading the file. Rather than reverting 75bd50fa841 ("drivers/base/node.c: use bin_attribute to break the size limitation of cpumap ABI") let's put in a size value at compile time. Use direct comparison and a worst-case maximum to ensure compile time constants. For cpulist the max is on the order of NR_CPUS * (ceil(log10(NR_CPUS)) + 1) which for 8192 is 40960 (8192 * 5). In order to get near that you'd need a system with every other CPU on one node or something similar. e.g. (0,2,4,... 1024,1026...). To simplify the math and support larger NR_CPUS we are using NR_CPUS * 6. We also set it to a min of PAGE_SIZE to retain the older behavior for smaller NR_CPUS. The cpumap file wants to be something like NR_CPUS/4 + NR_CPUS/32, for the ","s so for simplicity we are using NR_CPUS/2.

On an 80 cpu 4-node sytem (NR_CPUS == 8192)

before:

(quoted diff) +0 -2
-r--r--r--. 1 root root 0 Jul 12 14:08 /sys/devices/system/node/node0/cpulist
-r--r--r--. 1 root root 0 Jul 11 17:25 /sys/devices/system/node/node0/cpumap

after:

(quoted diff) +0 -2
-r--r--r--. 1 root root 49152 Jul 13 09:26 /sys/devices/system/node/node0/cpulist
-r--r--r--. 1 root root  4096 Jul 13 09:24 /sys/devices/system/node/node0/cpumap
Fixes 75bd50fa841 ("drivers/base/node.c: use bin_attribute to break the size limitation of cpumap ABI")
Cc Greg Kroah-Hartman <[email protected]>
Cc "Rafael J. Wysocki" <[email protected]>
---
 drivers/base/node.c | 4 ++--
 1 file changed, 2 insertions(+), 2 deletions(-)
drivers/base/node.c +0 -0
index 0ac6376ef7a1..3577f4eb4ac6 100644
@@ -45,7 +45,7 @@ static inline ssize_t cpumap_read(struct file *file, struct kobject *kobj,
 	return n;
 }
 
Quoted 5 lines

drivers/base/node.c static inline ssize_t cpumap_read(struct file *file, struct kobject *kobj, · line 48

Quoted 2 lines
drivers/base/node.c +1 -1
-static BIN_ATTR_RO(cpumap, 0);
+static BIN_ATTR_RO(cpumap, (((NR_CPUS>>1) > PAGE_SIZE) ? NR_CPUS >> 1 : PAGE_SIZE));

Why not just PAGE_SIZE?

Quoted 6 lines

This one is likely to grow beyond page size if NR_CPUS is larger than 8192. It's about 2300 bytes at 8192 if my math is right.

Let me ask again before I write the comments... what values of NR_CPUS are you seeing? Are people using an order of magnitude above 8192? Two orders? This patch (with the 6 below) handles about 1 order more.

One order for now that we know of, probably larger that we don't :)

Quoted 6 lines
Quoted 6 lines
Quoted 6 lines
drivers/base/node.c +0 -0
 static inline ssize_t cpulist_read(struct file *file, struct kobject *kobj,
 				   struct bin_attribute *attr, char *buf,
@@ -66,7 +66,7 @@ static inline ssize_t cpulist_read(struct file *file, struct kobject *kobj,
 	return n;
 }
 
Quoted 7 lines
Quoted 5 lines

drivers/base/node.c static inline ssize_t cpulist_read(struct file *file, struct kobject *kobj, · line 69

Quoted 2 lines
drivers/base/node.c +1 -1
-static BIN_ATTR_RO(cpulist, 0);
+static BIN_ATTR_RO(cpulist, (((NR_CPUS * 6) > PAGE_SIZE) ? NR_CPUS *6 : PAGE_SIZE));

"* 6" in both places here?

It's compile time so it's cheap. But yes, 6 in both places. That's to support NR_CPUS=100000.

Sorry, I was trying to point out to use a " " before the "6" in both places. Not questioning the math.

thanks,

greg k-h

author
#4 lore raw
Quoted 49 lines · On Wed, Jul 13, 2022 at 04:25:30PM +0200 Greg Kroah-Hartman wrote:
Quoted 48 lines · Phil Auld
Quoted 46 lines · On Wed, Jul 13, 2022 at 03:58:43PM +0200 Greg Kroah-Hartman wrote:
Quoted 37 lines · Phil Auld

Using bin_attributes with a 0 size causes fstat and friends to return that 0 size. This breaks userspace code that retrieves the size before reading the file. Rather than reverting 75bd50fa841 ("drivers/base/node.c: use bin_attribute to break the size limitation of cpumap ABI") let's put in a size value at compile time. Use direct comparison and a worst-case maximum to ensure compile time constants. For cpulist the max is on the order of NR_CPUS * (ceil(log10(NR_CPUS)) + 1) which for 8192 is 40960 (8192 * 5). In order to get near that you'd need a system with every other CPU on one node or something similar. e.g. (0,2,4,... 1024,1026...). To simplify the math and support larger NR_CPUS we are using NR_CPUS * 6. We also set it to a min of PAGE_SIZE to retain the older behavior for smaller NR_CPUS. The cpumap file wants to be something like NR_CPUS/4 + NR_CPUS/32, for the ","s so for simplicity we are using NR_CPUS/2.

On an 80 cpu 4-node sytem (NR_CPUS == 8192)

before:

(quoted diff) +0 -2
-r--r--r--. 1 root root 0 Jul 12 14:08 /sys/devices/system/node/node0/cpulist
-r--r--r--. 1 root root 0 Jul 11 17:25 /sys/devices/system/node/node0/cpumap

after:

(quoted diff) +0 -2
-r--r--r--. 1 root root 49152 Jul 13 09:26 /sys/devices/system/node/node0/cpulist
-r--r--r--. 1 root root  4096 Jul 13 09:24 /sys/devices/system/node/node0/cpumap
Fixes 75bd50fa841 ("drivers/base/node.c: use bin_attribute to break the size limitation of cpumap ABI")
Cc Greg Kroah-Hartman <[email protected]>
Cc "Rafael J. Wysocki" <[email protected]>
---
 drivers/base/node.c | 4 ++--
 1 file changed, 2 insertions(+), 2 deletions(-)
drivers/base/node.c +0 -0
index 0ac6376ef7a1..3577f4eb4ac6 100644

drivers/base/node.c static inline ssize_t cpumap_read(struct file *file, struct kobject *kobj, · line 48

Quoted 5 lines
drivers/base/node.c +1 -1
@@ -45,7 +45,7 @@ static inline ssize_t cpumap_read(struct file *file, struct kobject *kobj,
 	return n;
 }
 
-static BIN_ATTR_RO(cpumap, 0);
+static BIN_ATTR_RO(cpumap, (((NR_CPUS>>1) > PAGE_SIZE) ? NR_CPUS >> 1 : PAGE_SIZE));

Why not just PAGE_SIZE?

Quoted 8 lines
Quoted 6 lines

This one is likely to grow beyond page size if NR_CPUS is larger than 8192. It's about 2300 bytes at 8192 if my math is right.

Let me ask again before I write the comments... what values of NR_CPUS are you seeing? Are people using an order of magnitude above 8192? Two orders? This patch (with the 6 below) handles about 1 order more.

One order for now that we know of, probably larger that we don't :)

Fair enough. I'll go with 7 then. Thanks!

Quoted 13 lines
Quoted 13 lines
Quoted 12 lines

drivers/base/node.c static inline ssize_t cpulist_read(struct file *file, struct kobject *kobj, · line 69

Quoted 9 lines
drivers/base/node.c +1 -1
 static inline ssize_t cpulist_read(struct file *file, struct kobject *kobj,
 				   struct bin_attribute *attr, char *buf,
@@ -66,7 +66,7 @@ static inline ssize_t cpulist_read(struct file *file, struct kobject *kobj,
 	return n;
 }
 
-static BIN_ATTR_RO(cpulist, 0);
+static BIN_ATTR_RO(cpulist, (((NR_CPUS * 6) > PAGE_SIZE) ? NR_CPUS *6 : PAGE_SIZE));

"* 6" in both places here?

Quoted 5 lines
Quoted 1 line

It's compile time so it's cheap. But yes, 6 in both places. That's to support NR_CPUS=100000.

Sorry, I was trying to point out to use a " " before the "6" in both places. Not questioning the math.

Right! Sorry, misunderstood, obviously :)

Cheers,
Phil

Quoted 4 lines

thanks,

greg k-h