mm/hotplug: Enable runtime update of memmap_on_memory parameter

Commit Message

Aneesh Kumar K.V July 21, 2023, 1:19 p.m. UTC

Signed-off-by: Aneesh Kumar K.V <[e-mail]>
---
This is dependent on patches posted at
https://lore.kernel.org/linux-mm/[e-mail]/

 mm/memory_hotplug.c | 27 +++++++++++++++------------
 1 file changed, 15 insertions(+), 12 deletions(-)

Comments

Andrew Morton July 24, 2023, 5:31 p.m. UTC | #1

On Fri, 21 Jul 2023 18:49:50 +0530 "Aneesh Kumar K.V" <[e-mail]> wrote:

> Signed-off-by: Aneesh Kumar K.V <[e-mail]>
> ---
> This is dependent on patches posted at
> https://lore.kernel.org/linux-mm/[e-mail]/

It appears that the above-linked series is to be updated, so
would it be appropriate to append this patch to that series?

If not, please resend this once the above-linked series has landed
in mm-unstable, thanks.

David Hildenbrand July 24, 2023, 5:46 p.m. UTC | #2

On 24.07.23 19:31, Andrew Morton wrote:
> On Fri, 21 Jul 2023 18:49:50 +0530 "Aneesh Kumar K.V" <[e-mail]> wrote:
> 
>> Signed-off-by: Aneesh Kumar K.V <[e-mail]>
>> ---
>> This is dependent on patches posted at
>> https://lore.kernel.org/linux-mm/[e-mail]/
> 
> It appears that the above-linked series is to be updated, so
> would it be appropriate to append this patch to that series?
> 
> If not, please resend this once the above-linked series has landed
> in mm-unstable, thanks.
> 

Yes, let's include that patch in that series.

In general, LGTM. Toggling it at runtime, however, makes it harder for a 
driver to stabilize on the value. In the context of virtio-mem this 
would get important (once supporting memmap_on_memory).

Patch

diff --git a/mm/memory_hotplug.c b/mm/memory_hotplug.c
index 6a8adbe030f9..21a5113fe9be 100644
--- a/mm/memory_hotplug.c
+++ b/mm/memory_hotplug.c
@@ -86,7 +86,12 @@  static int set_memmap_mode(const char *val, const struct kernel_param *kp)
 		mode =  MEMMAP_ON_MEMORY_DISABLE;
 
 matched:
+	/*
+	 * Avoid changing memmap mode during hotplug.
+	 */
+	get_online_mems();
 	*((int *)kp->arg) =  mode;
+	put_online_mems();
 	if (mode == MEMMAP_ON_MEMORY_FORCE) {
 		pr_info("Memory hotplug will reserve %ld pages in each memory block\n",
 			memory_block_align_base(memory_block_size_bytes()));
@@ -108,7 +113,7 @@  static const struct kernel_param_ops memmap_mode_ops = {
 	.set = set_memmap_mode,
 	.get = get_memmap_mode,
 };
-module_param_cb(memmap_on_memory, &memmap_mode_ops, &memmap_mode, 0444);
+module_param_cb(memmap_on_memory, &memmap_mode_ops, &memmap_mode, 0644);
 MODULE_PARM_DESC(memmap_on_memory, "Enable memmap on memory for memory hotplug\n"
 	"With value \"force\" it could result in memory wastage due to memmap size limitations \n"
 	"For example, if the memmap for a memory block requires 1 MiB, but the pageblock \n"
@@ -2162,18 +2167,16 @@  static int __ref try_remove_memory(u64 start, u64 size)
 	 * We only support removing memory added with MHP_MEMMAP_ON_MEMORY in
 	 * the same granularity it was added - a single memory block.
 	 */
-	if (mhp_memmap_on_memory()) {
-		ret = walk_memory_blocks(start, size, &mhp_altmap,
-					 get_vmemmap_altmap_cb);
-		if (ret) {
-			if (size != memory_block_size_bytes()) {
-				pr_warn("Refuse to remove %#llx - %#llx,"
-					"wrong granularity\n",
-					start, start + size);
-				return -EINVAL;
-			}
-			altmap = &mhp_altmap;
+	ret = walk_memory_blocks(start, size, &mhp_altmap,
+				 get_vmemmap_altmap_cb);
+	if (ret) {
+		if (size != memory_block_size_bytes()) {
+			pr_warn("Refuse to remove %#llx - %#llx,"
+				"wrong granularity\n",
+				start, start + size);
+			return -EINVAL;
 		}
+		altmap = &mhp_altmap;
 	}
 
 	/* remove memmap entry */