From: lantianyu1986@gmail.com
To: kys@microsoft.com, haiyangz@microsoft.com,
sthemmin@microsoft.com, sashal@kernel.org,
michael.h.kelley@microsoft.com, david@redhat.com
Cc: Tianyu Lan <Tianyu.Lan@microsoft.com>,
linux-hyperv@vger.kernel.org, linux-kernel@vger.kernel.org,
vkuznets@redhat.com, eric.devolder@oracle.com
Subject: [RFC PATCH V2 5/10] x86/Hyper-V/Balloon: Avoid releasing ha_lock when traverse ha_region_list
Date: Tue, 7 Jan 2020 21:09:45 +0800 [thread overview]
Message-ID: <20200107130950.2983-6-Tianyu.Lan@microsoft.com> (raw)
In-Reply-To: <20200107130950.2983-1-Tianyu.Lan@microsoft.com>
From: Tianyu Lan <Tianyu.Lan@microsoft.com>
ha_lock is to protect ha_region_list. It is held in
hv_online_page() and handle_pg_range(). handle_pg_range()
is to traverse ha region list, find associated hot-add region
and add memory into system. hv_online_page() is called inside
of add_memory(). Current code is to release ha_lock before
calling add_memory() to avoid holding ha_lock twice in the
hv_online_page().
To avoid releasing ha_lock, add "lock_thread" in the struct hv_
dynmem_device to record thread of traversing ha region list,
check "lock_thread" in the hv_online_page() and try holding
ha_lock when current thread is not "lock_thread".
Signed-off-by: Tianyu Lan <Tianyu.Lan@microsoft.com>
---
drivers/hv/hv_balloon.c | 20 ++++++++++++--------
1 file changed, 12 insertions(+), 8 deletions(-)
diff --git a/drivers/hv/hv_balloon.c b/drivers/hv/hv_balloon.c
index 185146795122..729dc5551302 100644
--- a/drivers/hv/hv_balloon.c
+++ b/drivers/hv/hv_balloon.c
@@ -547,6 +547,7 @@ struct hv_dynmem_device {
* regions from ha_region_list.
*/
struct mutex ha_lock;
+ struct task_struct *lock_thread;
/*
* A list of hot-add regions.
@@ -711,7 +712,6 @@ static void hv_mem_hot_add(unsigned long start, unsigned long size,
for (i = 0; i < (size/HA_CHUNK); i++) {
start_pfn = start + (i * HA_CHUNK);
- mutex_lock(&dm_device.ha_lock);
has->ha_end_pfn += HA_CHUNK;
if (total_pfn > HA_CHUNK) {
@@ -723,7 +723,6 @@ static void hv_mem_hot_add(unsigned long start, unsigned long size,
}
has->covered_end_pfn += processed_pfn;
- mutex_unlock(&dm_device.ha_lock);
init_completion(&dm_device.ol_waitevent);
dm_device.ha_waiting = !memhp_auto_online;
@@ -744,10 +743,8 @@ static void hv_mem_hot_add(unsigned long start, unsigned long size,
*/
do_hot_add = false;
}
- mutex_lock(&dm_device.ha_lock);
has->ha_end_pfn -= HA_CHUNK;
has->covered_end_pfn -= processed_pfn;
- mutex_unlock(&dm_device.ha_lock);
break;
}
@@ -769,8 +766,14 @@ static void hv_online_page(struct page *pg, unsigned int order)
{
struct hv_hotadd_state *has;
unsigned long pfn = page_to_pfn(pg);
+ int need_unlock;
+
+ /* If current thread hasn't hold ha_lock, take ha_lock here. */
+ if (dm_device.lock_thread != current) {
+ mutex_lock(&dm_device.ha_lock);
+ need_unlock = 1;
+ }
- mutex_lock(&dm_device.ha_lock);
list_for_each_entry(has, &dm_device.ha_region_list, list) {
/* The page belongs to a different HAS. */
if ((pfn < has->start_pfn) ||
@@ -780,7 +783,8 @@ static void hv_online_page(struct page *pg, unsigned int order)
hv_bring_pgs_online(has, pfn, 1UL << order);
break;
}
- mutex_unlock(&dm_device.ha_lock);
+ if (need_unlock)
+ mutex_unlock(&dm_device.ha_lock);
}
static int pfn_covered(unsigned long start_pfn, unsigned long pfn_cnt)
@@ -857,6 +861,7 @@ static unsigned long handle_pg_range(unsigned long pg_start,
pg_start);
mutex_lock(&dm_device.ha_lock);
+ dm_device.lock_thread = current;
list_for_each_entry(has, &dm_device.ha_region_list, list) {
/*
* If the pfn range we are dealing with is not in the current
@@ -909,9 +914,7 @@ static unsigned long handle_pg_range(unsigned long pg_start,
} else {
pfn_cnt = size;
}
- mutex_unlock(&dm_device.ha_lock);
hv_mem_hot_add(has->ha_end_pfn, size, pfn_cnt, has);
- mutex_lock(&dm_device.ha_lock);
}
/*
* If we managed to online any pages that were given to us,
@@ -920,6 +923,7 @@ static unsigned long handle_pg_range(unsigned long pg_start,
res = has->covered_end_pfn - old_covered_state;
break;
}
+ dm_device.lock_thread = NULL;
mutex_unlock(&dm_device.ha_lock);
return res;
--
2.14.5
next prev parent reply other threads:[~2020-01-07 13:10 UTC|newest]
Thread overview: 21+ messages / expand[flat|nested] mbox.gz Atom feed top
2020-01-07 13:09 [RFC PATCH V2 00/10] x86/Hyper-V: Add Dynamic memory hot-remove function lantianyu1986
2020-01-07 13:09 ` [RFC PATCH V2 1/10] mm/resource: Move child to new resource when release mem region lantianyu1986
2020-01-20 18:34 ` Michael Kelley
2020-01-20 19:20 ` Michael Kelley
2020-01-07 13:09 ` [RFC PATCH V2 2/10] mm: expose is_mem_section_removable() symbol lantianyu1986
2020-01-07 13:36 ` Michal Hocko
2020-01-10 13:41 ` David Hildenbrand
2020-01-13 14:49 ` [EXTERNAL] " Tianyu Lan
2020-01-13 15:01 ` David Hildenbrand
2020-01-14 9:50 ` Michal Hocko
2020-01-17 16:35 ` Tianyu Lan
2020-01-20 14:14 ` Michal Hocko
2020-01-07 13:09 ` [RFC PATCH V2 3/10] x86/Hyper-V/Balloon: Replace hot-add and balloon up works with a common work lantianyu1986
2020-01-20 19:12 ` Michael Kelley
2020-01-07 13:09 ` [RFC PATCH V2 4/10] x86/Hyper-V/Balloon: Convert spin lock ha_lock to mutex lantianyu1986
2020-01-07 13:09 ` lantianyu1986 [this message]
2020-01-07 13:09 ` [RFC PATCH V2 6/10] x86/Hyper-V/Balloon: Enable mem hot-remove capability lantianyu1986
2020-01-07 13:09 ` [RFC PATCH V2 7/10] x86/Hyper-V/Balloon: Handle mem hot-remove request lantianyu1986
2020-01-07 13:09 ` [RFC PATCH V2 8/10] x86/Hyper-V/Balloon: Handle request with non-aligned page number lantianyu1986
2020-01-07 13:09 ` [RFC PATCH V2 9/10] x86/Hyper-V/Balloon: Hot add mem in the gaps of hot add region lantianyu1986
2020-01-07 13:09 ` [RFC PATCH V2 10/10] x86/Hyper-V: Workaround Hyper-V unballoon msg bug lantianyu1986
Reply instructions:
You may reply publicly to this message via plain-text email
using any one of the following methods:
* Save the following mbox file, import it into your mail client,
and reply-to-all from there: mbox
Avoid top-posting and favor interleaved quoting:
https://en.wikipedia.org/wiki/Posting_style#Interleaved_style
* Reply using the --to, --cc, and --in-reply-to
switches of git-send-email(1):
git send-email \
--in-reply-to=20200107130950.2983-6-Tianyu.Lan@microsoft.com \
--to=lantianyu1986@gmail.com \
--cc=Tianyu.Lan@microsoft.com \
--cc=david@redhat.com \
--cc=eric.devolder@oracle.com \
--cc=haiyangz@microsoft.com \
--cc=kys@microsoft.com \
--cc=linux-hyperv@vger.kernel.org \
--cc=linux-kernel@vger.kernel.org \
--cc=michael.h.kelley@microsoft.com \
--cc=sashal@kernel.org \
--cc=sthemmin@microsoft.com \
--cc=vkuznets@redhat.com \
/path/to/YOUR_REPLY
https://kernel.org/pub/software/scm/git/docs/git-send-email.html
* If your mail client supports setting the In-Reply-To header
via mailto: links, try the mailto: link
Be sure your reply has a Subject: header at the top and a blank line
before the message body.
This is a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox;
as well as URLs for NNTP newsgroup(s).