Skip to contentSkip to conversation

x86/uv/time: Replace one-element array and save heap space

PATCH Applied Gustavo A. R. Silva 4 participants 4 messages Oct 1, 2020

Subsystems (3)
lore
Share & tools
lore
#0 lore raw

PATCH v2 1/1 x86/uv/time: Replace one-element array and save heap space

There is a regular need in the kernel to provide a way to declare having a dynamically sized set of trailing elements in a structure. Kernel code should always use “flexible array members”[1] for these cases. The older style of one-element or zero-length arrays should no longer be used[2].

struct uv_rtc_timer_head contains a one-element array cpu[1].

Switch it to a flexible array and use the struct_size() helper to calculate the allocation size. Also, save some heap space in the process[3].

[1] https://en.wikipedia.org/wiki/Flexible_array_member
[2] https://www.kernel.org/doc/html/v5.9-rc1/process/deprecated.html#zero-length-and-one-element-arrays
[3] https://lore.kernel.org/lkml/20200518190114.GA7757@embeddedor/

---
Changes in v2:
- Modify changelog text (Thomas).
- Add Reviewed-by and Build-tested-by tags.
- Add a public link to the build-test results.

 arch/x86/platform/uv/uv_time.c | 7 +++----
 1 file changed, 3 insertions(+), 4 deletions(-)
arch/x86/platform/uv/uv_time.c +3 -4
index f82a1337a608..6c348c2d0def 100644
@@ -52,7 +52,7 @@ struct uv_rtc_timer_head {
 	struct {
 		int	lcpu;		/* systemwide logical cpu number */
 		u64	expires;	/* next timer expiration for this cpu */
-	} cpu[1];
+	} cpu[];
 };
 
 /*
@@ -148,9 +148,8 @@ static __init int uv_rtc_allocate_timers(void)
 		struct uv_rtc_timer_head *head = blade_info[bid];
 
 		if (!head) {
-			head = kmalloc_node(sizeof(struct uv_rtc_timer_head) +
-				(uv_blade_nr_possible_cpus(bid) *
-					2 * sizeof(u64)),
+			head = kmalloc_node(struct_size(head, cpu,
+				uv_blade_nr_possible_cpus(bid)),
 				GFP_KERNEL, nid);
 			if (!head) {
 				uv_rtc_deallocate_timers();
Signature · 1 line
2.27.0
#1 lore raw
Quoted 10 lines · Gustavo A. R. Silva

There is a regular need in the kernel to provide a way to declare having a dynamically sized set of trailing elements in a structure. Kernel code should always use “flexible array members”[1] for these cases. The older style of one-element or zero-length arrays should no longer be used[2].

struct uv_rtc_timer_head contains a one-element array cpu[1].

Switch it to a flexible array and use the struct_size() helper to calculate the allocation size. Also, save some heap space in the process[3].

trivia:

arch/x86/platform/uv/uv_time.c

Quoted 1 line
arch/x86/platform/uv/uv_time.c +0 -0

[]

arch/x86/platform/uv/uv_time.c static __init int uv_rtc_allocate_timers(void) · line 155

Quoted 12 lines
arch/x86/platform/uv/uv_time.c +2 -3
@@ -148,9 +148,8 @@ static __init int uv_rtc_allocate_timers(void)
 		struct uv_rtc_timer_head *head = blade_info[bid];
 
 		if (!head) {
-			head = kmalloc_node(sizeof(struct uv_rtc_timer_head) +
-				(uv_blade_nr_possible_cpus(bid) *
-					2 * sizeof(u64)),
+			head = kmalloc_node(struct_size(head, cpu,
+				uv_blade_nr_possible_cpus(bid)),
 				GFP_KERNEL, nid);
 			if (!head) {
 				uv_rtc_deallocate_timers();

Maybe save the value of uv_blade_nr_possible_cpus(bid) to reduce duplication and make the sizeof_struct more readable?

		if (!head) {
			int ncpus = uv_blade_nr_possible_cpus(bid);

			head = kmalloc_node(struct_size(head, cpu, ncpus),
					    GFP_KERNEL, nid);
			if (!head) {
				uv_rtc_deallocate_timers();
				return -ENOMEM;
			}
			spin_lock_init(&head->lock);
			head->ncpus = ncpus;
			head->next_cpu = -1;
			blade_info[bid] = head;
		}
#2 lore raw

[tip: x86/platform] x86/uv/time: Use a flexible array in struct uv_rtc_timer_head

In reply to Gustavo A. R. Silva
The following commit has been merged into the x86/platform branch of tip:

The following commit has been merged into the x86/platform branch of tip:

Commit-ID:     a0947081af2ac9549e6ba19877456730713bde23
Gitweb:        https://git.kernel.org/tip/a0947081af2ac9549e6ba19877456730713bde23
Author:        Gustavo A. R. Silva <[email protected]>
AuthorDate:    Thu, 01 Oct 2020 09:56:08 -05:00
Committer:     Borislav Petkov <[email protected]>
CommitterDate: Thu, 01 Oct 2020 18:47:39 +02:00

x86/uv/time: Use a flexible array in struct uv_rtc_timer_head

There is a regular need in the kernel to provide a way to declare having a dynamically sized set of trailing elements in a structure. Kernel code should always use “flexible array members”[1] for these cases. The older style of one-element or zero-length arrays should no longer be used[2].

struct uv_rtc_timer_head contains a one-element array cpu[1]. Switch it to a flexible array and use the struct_size() helper to calculate the allocation size. Also, save some heap space in the process[3].

[1] https://en.wikipedia.org/wiki/Flexible_array_member
[2] https://www.kernel.org/doc/html/v5.9-rc1/process/deprecated.html#zero-length-and-one-element-arrays
[3] https://lore.kernel.org/lkml/20200518190114.GA7757@embeddedor/

[ bp: Massage a bit. ]

---
 arch/x86/platform/uv/uv_time.c | 7 +++----
 1 file changed, 3 insertions(+), 4 deletions(-)
arch/x86/platform/uv/uv_time.c +3 -4
index f82a133..6c348c2 100644
@@ -52,7 +52,7 @@ struct uv_rtc_timer_head {
 	struct {
 		int	lcpu;		/* systemwide logical cpu number */
 		u64	expires;	/* next timer expiration for this cpu */
-	} cpu[1];
+	} cpu[];
 };
 
 /*
@@ -148,9 +148,8 @@ static __init int uv_rtc_allocate_timers(void)
 		struct uv_rtc_timer_head *head = blade_info[bid];
 
 		if (!head) {
-			head = kmalloc_node(sizeof(struct uv_rtc_timer_head) +
-				(uv_blade_nr_possible_cpus(bid) *
-					2 * sizeof(u64)),
+			head = kmalloc_node(struct_size(head, cpu,
+				uv_blade_nr_possible_cpus(bid)),
 				GFP_KERNEL, nid);
 			if (!head) {
 				uv_rtc_deallocate_timers();
maintainer Reviewed-by on v2 1/1
#3 lore raw
In reply to Gustavo A. R. Silva
Quoted 4 lines · Gustavo A. R. Silva

There is a regular need in the kernel to provide a way to declare having a dynamically sized set of trailing elements in a structure. Kernel code should always use “flexible array members”[1] for these cases. The older style of one-element or zero-length arrays should no longer be used[2].

I see this has already gone in, so I'm late. But for whatever it's worth, it's a good change. Thank you.

--> Steve wahl

Quoted 51 lines

struct uv_rtc_timer_head contains a one-element array cpu[1].

Switch it to a flexible array and use the struct_size() helper to calculate the allocation size. Also, save some heap space in the process[3].

[1] https://en.wikipedia.org/wiki/Flexible_array_member
[2] https://www.kernel.org/doc/html/v5.9-rc1/process/deprecated.html#zero-length-and-one-element-arrays
[3] https://lore.kernel.org/lkml/20200518190114.GA7757@embeddedor/

---
Changes in v2:
- Modify changelog text (Thomas).
- Add Reviewed-by and Build-tested-by tags.
- Add a public link to the build-test results.

 arch/x86/platform/uv/uv_time.c | 7 +++----
 1 file changed, 3 insertions(+), 4 deletions(-)
arch/x86/platform/uv/uv_time.c +3 -4
index f82a1337a608..6c348c2d0def 100644
@@ -52,7 +52,7 @@ struct uv_rtc_timer_head {
 	struct {
 		int	lcpu;		/* systemwide logical cpu number */
 		u64	expires;	/* next timer expiration for this cpu */
-	} cpu[1];
+	} cpu[];
 };
 
 /*
@@ -148,9 +148,8 @@ static __init int uv_rtc_allocate_timers(void)
 		struct uv_rtc_timer_head *head = blade_info[bid];
 
 		if (!head) {
-			head = kmalloc_node(sizeof(struct uv_rtc_timer_head) +
-				(uv_blade_nr_possible_cpus(bid) *
-					2 * sizeof(u64)),
+			head = kmalloc_node(struct_size(head, cpu,
+				uv_blade_nr_possible_cpus(bid)),
 				GFP_KERNEL, nid);
 			if (!head) {
 				uv_rtc_deallocate_timers();
Signature · 1 line
2.27.0