From: Chao Yu <chao@kernel.org>
To: Sandeep Dhavale <dhavale@google.com>,
linux-erofs@lists.ozlabs.org, Gao Xiang <xiang@kernel.org>,
Yue Hu <zbestahu@gmail.com>,
Jeffle Xu <jefflexu@linux.alibaba.com>
Cc: chao@kernel.org, hsiangkao@linux.alibaba.com,
kernel-team@android.com, linux-kernel@vger.kernel.org
Subject: Re: [PATCH v4] erofs: lazily initialize per-CPU workers and CPU hotplug hooks
Date: Fri, 25 Apr 2025 10:15:31 +0800 [thread overview]
Message-ID: <894ca2d3-e680-4395-9887-2b6060fc8096@kernel.org> (raw)
In-Reply-To: <20250423061023.131354-1-dhavale@google.com>
On 4/23/25 14:10, Sandeep Dhavale wrote:
> Currently, when EROFS is built with per-CPU workers, the workers are
> started and CPU hotplug hooks are registered during module initialization.
> This leads to unnecessary worker start/stop cycles during CPU hotplug
> events, particularly on Android devices that frequently suspend and resume.
>
> This change defers the initialization of per-CPU workers and the
> registration of CPU hotplug hooks until the first EROFS mount. This
> ensures that these resources are only allocated and managed when EROFS is
> actually in use.
>
> The tear down of per-CPU workers and unregistration of CPU hotplug hooks
> still occurs during z_erofs_exit_subsystem(), but only if they were
> initialized.
>
> Signed-off-by: Sandeep Dhavale <dhavale@google.com>
> ---
> v3: https://lore.kernel.org/linux-erofs/20250422234546.2932092-1-dhavale@google.com/
> Changes since v3:
> - fold z_erofs_init_pcpu_workers() in the caller and rename the caller
>
> fs/erofs/zdata.c | 61 +++++++++++++++++++++++++++++++++++-------------
> 1 file changed, 45 insertions(+), 16 deletions(-)
>
> diff --git a/fs/erofs/zdata.c b/fs/erofs/zdata.c
> index 0671184d9cf1..647a8340c9a1 100644
> --- a/fs/erofs/zdata.c
> +++ b/fs/erofs/zdata.c
> @@ -291,6 +291,9 @@ static struct workqueue_struct *z_erofs_workqueue __read_mostly;
>
> #ifdef CONFIG_EROFS_FS_PCPU_KTHREAD
> static struct kthread_worker __rcu **z_erofs_pcpu_workers;
> +static atomic_t erofs_percpu_workers_initialized = ATOMIC_INIT(0);
> +static int erofs_cpu_hotplug_init(void);
> +static void erofs_cpu_hotplug_destroy(void);
>
> static void erofs_destroy_percpu_workers(void)
> {
> @@ -336,9 +339,40 @@ static int erofs_init_percpu_workers(void)
> }
> return 0;
> }
> +
> +static int z_erofs_init_pcpu_workers(void)
> +{
> + int err;
> +
> + if (atomic_xchg(&erofs_percpu_workers_initialized, 1))
> + return 0;
> +
> + err = erofs_init_percpu_workers();
> + if (err)
> + goto err_init_percpu_workers;
> +
> + err = erofs_cpu_hotplug_init();
> + if (err < 0)
> + goto err_cpuhp_init;
> + return err;
> +
> +err_cpuhp_init:
> + erofs_destroy_percpu_workers();
> +err_init_percpu_workers:
> + atomic_set(&erofs_percpu_workers_initialized, 0);
> + return err;
> +}
- mount #1 - mount #2
- z_erofs_init_pcpu_workers
- atomic_xchg(, 1)
- z_erofs_init_pcpu_workers
- atomic_xchg(, 1)
: return 0 since atomic variable is 1
it will run w/o percpu workers and hotplug
: update atomic variable to 1
- erofs_init_percpu_workers
: fail
- atomic_set(, 0)
: update atomic variable to 0 & fail the mount
Can we add some logs to show we succeed/fail to initialize workers or
hotplugs? As for mount #2, it expects it will run w/ them, but finally
it may not. So we'd better have a simple way to know?
Thanks,
> +
> +static void z_erofs_destroy_pcpu_workers(void)
> +{
> + if (!atomic_xchg(&erofs_percpu_workers_initialized, 0))
> + return;
> + erofs_cpu_hotplug_destroy();
> + erofs_destroy_percpu_workers();
> +}
> #else
> -static inline void erofs_destroy_percpu_workers(void) {}
> -static inline int erofs_init_percpu_workers(void) { return 0; }
> +static inline int z_erofs_init_pcpu_workers(void) { return 0; }
> +static inline void z_erofs_destroy_pcpu_workers(void) {}
> #endif
>
> #if defined(CONFIG_HOTPLUG_CPU) && defined(CONFIG_EROFS_FS_PCPU_KTHREAD)
> @@ -405,8 +439,7 @@ static inline void erofs_cpu_hotplug_destroy(void) {}
>
> void z_erofs_exit_subsystem(void)
> {
> - erofs_cpu_hotplug_destroy();
> - erofs_destroy_percpu_workers();
> + z_erofs_destroy_pcpu_workers();
> destroy_workqueue(z_erofs_workqueue);
> z_erofs_destroy_pcluster_pool();
> z_erofs_exit_decompressor();
> @@ -430,19 +463,8 @@ int __init z_erofs_init_subsystem(void)
> goto err_workqueue_init;
> }
>
> - err = erofs_init_percpu_workers();
> - if (err)
> - goto err_pcpu_worker;
> -
> - err = erofs_cpu_hotplug_init();
> - if (err < 0)
> - goto err_cpuhp_init;
> return err;
>
> -err_cpuhp_init:
> - erofs_destroy_percpu_workers();
> -err_pcpu_worker:
> - destroy_workqueue(z_erofs_workqueue);
> err_workqueue_init:
> z_erofs_destroy_pcluster_pool();
> err_pcluster_pool:
> @@ -644,10 +666,17 @@ static const struct address_space_operations z_erofs_cache_aops = {
>
> int z_erofs_init_super(struct super_block *sb)
> {
> - struct inode *const inode = new_inode(sb);
> + struct inode *inode;
> + int err;
>
> + err = z_erofs_init_pcpu_workers();
> + if (err)
> + return err;
> +
> + inode = new_inode(sb);
> if (!inode)
> return -ENOMEM;
> +
> set_nlink(inode, 1);
> inode->i_size = OFFSET_MAX;
> inode->i_mapping->a_ops = &z_erofs_cache_aops;
next prev parent reply other threads:[~2025-04-25 2:15 UTC|newest]
Thread overview: 7+ messages / expand[flat|nested] mbox.gz Atom feed top
2025-04-23 6:10 [PATCH v4] erofs: lazily initialize per-CPU workers and CPU hotplug hooks Sandeep Dhavale
2025-04-24 0:31 ` Gao Xiang
2025-04-24 6:04 ` Sandeep Dhavale
2025-04-24 12:02 ` Gao Xiang
2025-04-25 2:15 ` Chao Yu [this message]
2025-04-28 21:49 ` Sandeep Dhavale
2025-04-29 6:05 ` Chao Yu
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=894ca2d3-e680-4395-9887-2b6060fc8096@kernel.org \
--to=chao@kernel.org \
--cc=dhavale@google.com \
--cc=hsiangkao@linux.alibaba.com \
--cc=jefflexu@linux.alibaba.com \
--cc=kernel-team@android.com \
--cc=linux-erofs@lists.ozlabs.org \
--cc=linux-kernel@vger.kernel.org \
--cc=xiang@kernel.org \
--cc=zbestahu@gmail.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