All of lore.kernel.org
 help / color / mirror / Atom feed
From: Todd Kjos <tkjos@google.com>
To: Li Li <dualli@chromium.org>
Cc: dualli@google.com, gregkh@linuxfoundation.org,
	christian@brauner.io, arve@android.com,
	devel@driverdev.osuosl.org, linux-kernel@vger.kernel.org,
	maco@google.com, hridya@google.com, surenb@google.com,
	joel@joelfernandes.org, kernel-team@android.com
Subject: Re: [RESEND PATCH v3 1/1] Binder: add TF_UPDATE_TXN to replace outdated txn
Date: Tue, 21 Jun 2022 10:11:07 -0700	[thread overview]
Message-ID: <CAHRSSEwJhpmiJtnK4j76uDh+EGXr9FwMotD8=vwr70EmVTfWsQ@mail.gmail.com> (raw)
In-Reply-To: <20220526220018.3334775-2-dualli@chromium.org>

On Thu, May 26, 2022 at 3:00 PM Li Li <dualli@chromium.org> wrote:
>
> From: Li Li <dualli@google.com>
>
> When the target process is busy, incoming oneway transactions are
> queued in the async_todo list. If the clients continue sending extra
> oneway transactions while the target process is frozen, this queue can
> become too large to accommodate new transactions. That's why binder
> driver introduced ONEWAY_SPAM_DETECTION to detect this situation. It's
> helpful to debug the async binder buffer exhausting issue, but the
> issue itself isn't solved directly.
>
> In real cases applications are designed to send oneway transactions
> repeatedly, delivering updated inforamtion to the target process.
> Typical examples are Wi-Fi signal strength and some real time sensor
> data. Even if the apps might only care about the lastet information,
> all outdated oneway transactions are still accumulated there until the
> frozen process is thawed later. For this kind of situations, there's
> no existing method to skip those outdated transactions and deliver the
> latest one only.
>
> This patch introduces a new transaction flag TF_UPDATE_TXN. To use it,
> use apps can set this new flag along with TF_ONE_WAY. When such an
> oneway transaction is to be queued into the async_todo list of a frozen
> process, binder driver will check if any previous pending transactions
> can be superseded by comparing their code, flags and target node. If
> such an outdated pending transaction is found, the latest transaction
> will supersede that outdated one. This effectively prevents the async
> binder buffer running out and saves unnecessary binder read workloads.
>
> Signed-off-by: Li Li <dualli@google.com>

Acked-by: Todd Kjos <tkjos@google.com>

> ---
> v3:
>   - Add this changelog required by "The canonical patch format"
> v2:
>   - Fix alignment warnings reported by checkpatch --strict
>   - Add descriptive summary in patch subject
>
>  drivers/android/binder.c            | 85 ++++++++++++++++++++++++++++-
>  drivers/android/binder_trace.h      |  4 ++
>  include/uapi/linux/android/binder.h |  1 +
>  3 files changed, 87 insertions(+), 3 deletions(-)
>
> diff --git a/drivers/android/binder.c b/drivers/android/binder.c
> index f3b639e89dd8..bb968cf2f9ec 100644
> --- a/drivers/android/binder.c
> +++ b/drivers/android/binder.c
> @@ -2594,6 +2594,56 @@ static int binder_fixup_parent(struct list_head *pf_head,
>         return binder_add_fixup(pf_head, buffer_offset, bp->buffer, 0);
>  }
>
> +/**
> + * binder_can_update_transaction() - Can a txn be superseded by an updated one?
> + * @t1: the pending async txn in the frozen process
> + * @t2: the new async txn to supersede the outdated pending one
> + *
> + * Return:  true if t2 can supersede t1
> + *          false if t2 can not supersede t1
> + */
> +static bool binder_can_update_transaction(struct binder_transaction *t1,
> +                                         struct binder_transaction *t2)
> +{
> +       if ((t1->flags & t2->flags & (TF_ONE_WAY | TF_UPDATE_TXN)) !=
> +           (TF_ONE_WAY | TF_UPDATE_TXN) || !t1->to_proc || !t2->to_proc)
> +               return false;
> +       if (t1->to_proc->tsk == t2->to_proc->tsk && t1->code == t2->code &&
> +           t1->flags == t2->flags && t1->buffer->pid == t2->buffer->pid &&
> +           t1->buffer->target_node->ptr == t2->buffer->target_node->ptr &&
> +           t1->buffer->target_node->cookie == t2->buffer->target_node->cookie)
> +               return true;
> +       return false;
> +}
> +
> +/**
> + * binder_find_outdated_transaction_ilocked() - Find the outdated transaction
> + * @t:          new async transaction
> + * @target_list: list to find outdated transaction
> + *
> + * Return: the outdated transaction if found
> + *         NULL if no outdated transacton can be found
> + *
> + * Requires the proc->inner_lock to be held.
> + */
> +static struct binder_transaction *
> +binder_find_outdated_transaction_ilocked(struct binder_transaction *t,
> +                                        struct list_head *target_list)
> +{
> +       struct binder_work *w;
> +
> +       list_for_each_entry(w, target_list, entry) {
> +               struct binder_transaction *t_queued;
> +
> +               if (w->type != BINDER_WORK_TRANSACTION)
> +                       continue;
> +               t_queued = container_of(w, struct binder_transaction, work);
> +               if (binder_can_update_transaction(t_queued, t))
> +                       return t_queued;
> +       }
> +       return NULL;
> +}
> +
>  /**
>   * binder_proc_transaction() - sends a transaction to a process and wakes it up
>   * @t:         transaction to send
> @@ -2619,6 +2669,7 @@ static int binder_proc_transaction(struct binder_transaction *t,
>         struct binder_node *node = t->buffer->target_node;
>         bool oneway = !!(t->flags & TF_ONE_WAY);
>         bool pending_async = false;
> +       struct binder_transaction *t_outdated = NULL;
>
>         BUG_ON(!node);
>         binder_node_lock(node);
> @@ -2646,12 +2697,24 @@ static int binder_proc_transaction(struct binder_transaction *t,
>         if (!thread && !pending_async)
>                 thread = binder_select_thread_ilocked(proc);
>
> -       if (thread)
> +       if (thread) {
>                 binder_enqueue_thread_work_ilocked(thread, &t->work);
> -       else if (!pending_async)
> +       } else if (!pending_async) {
>                 binder_enqueue_work_ilocked(&t->work, &proc->todo);
> -       else
> +       } else {
> +               if ((t->flags & TF_UPDATE_TXN) && proc->is_frozen) {
> +                       t_outdated = binder_find_outdated_transaction_ilocked(t,
> +                                                                             &node->async_todo);
> +                       if (t_outdated) {
> +                               binder_debug(BINDER_DEBUG_TRANSACTION,
> +                                            "txn %d supersedes %d\n",
> +                                            t->debug_id, t_outdated->debug_id);
> +                               list_del_init(&t_outdated->work.entry);
> +                               proc->outstanding_txns--;
> +                       }
> +               }
>                 binder_enqueue_work_ilocked(&t->work, &node->async_todo);
> +       }
>
>         if (!pending_async)
>                 binder_wakeup_thread_ilocked(proc, thread, !oneway /* sync */);
> @@ -2660,6 +2723,22 @@ static int binder_proc_transaction(struct binder_transaction *t,
>         binder_inner_proc_unlock(proc);
>         binder_node_unlock(node);
>
> +       /*
> +        * To reduce potential contention, free the outdated transaction and
> +        * buffer after releasing the locks.
> +        */
> +       if (t_outdated) {
> +               struct binder_buffer *buffer = t_outdated->buffer;
> +
> +               t_outdated->buffer = NULL;
> +               buffer->transaction = NULL;
> +               trace_binder_transaction_update_buffer_release(buffer);
> +               binder_transaction_buffer_release(proc, NULL, buffer, 0, 0);
> +               binder_alloc_free_buf(&proc->alloc, buffer);
> +               kfree(t_outdated);
> +               binder_stats_deleted(BINDER_STAT_TRANSACTION);
> +       }
> +
>         return 0;
>  }
>
> diff --git a/drivers/android/binder_trace.h b/drivers/android/binder_trace.h
> index 8eeccdc64724..8cc07e6a4273 100644
> --- a/drivers/android/binder_trace.h
> +++ b/drivers/android/binder_trace.h
> @@ -311,6 +311,10 @@ DEFINE_EVENT(binder_buffer_class, binder_transaction_failed_buffer_release,
>         TP_PROTO(struct binder_buffer *buffer),
>         TP_ARGS(buffer));
>
> +DEFINE_EVENT(binder_buffer_class, binder_transaction_update_buffer_release,
> +            TP_PROTO(struct binder_buffer *buffer),
> +            TP_ARGS(buffer));
> +
>  TRACE_EVENT(binder_update_page_range,
>         TP_PROTO(struct binder_alloc *alloc, bool allocate,
>                  void __user *start, void __user *end),
> diff --git a/include/uapi/linux/android/binder.h b/include/uapi/linux/android/binder.h
> index 11157fae8a8e..0b291b1d29d1 100644
> --- a/include/uapi/linux/android/binder.h
> +++ b/include/uapi/linux/android/binder.h
> @@ -271,6 +271,7 @@ enum transaction_flags {
>         TF_STATUS_CODE  = 0x08, /* contents are a 32-bit status code */
>         TF_ACCEPT_FDS   = 0x10, /* allow replies with file descriptors */
>         TF_CLEAR_BUF    = 0x20, /* clear buffer on txn complete */
> +       TF_UPDATE_TXN   = 0x40, /* update the outdated pending async txn */
>  };
>
>  struct binder_transaction_data {
> --
> 2.36.1.124.g0e6072fb45-goog
>

  reply	other threads:[~2022-06-21 17:11 UTC|newest]

Thread overview: 5+ messages / expand[flat|nested]  mbox.gz  Atom feed  top
2022-05-26 22:00 [RESEND PATCH v3 0/1] Binder: add TF_UPDATE_TXN to replace outdated txn Li Li
2022-05-26 22:00 ` [RESEND PATCH v3 1/1] " Li Li
2022-06-21 17:11   ` Todd Kjos [this message]
     [not found] ` <YpBmmVa+09ZhP9LH@kroah.com>
2022-06-15 18:05   ` [RESEND PATCH v3 0/1] " Li Li
2022-06-16  5:58     ` Greg KH

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='CAHRSSEwJhpmiJtnK4j76uDh+EGXr9FwMotD8=vwr70EmVTfWsQ@mail.gmail.com' \
    --to=tkjos@google.com \
    --cc=arve@android.com \
    --cc=christian@brauner.io \
    --cc=devel@driverdev.osuosl.org \
    --cc=dualli@chromium.org \
    --cc=dualli@google.com \
    --cc=gregkh@linuxfoundation.org \
    --cc=hridya@google.com \
    --cc=joel@joelfernandes.org \
    --cc=kernel-team@android.com \
    --cc=linux-kernel@vger.kernel.org \
    --cc=maco@google.com \
    --cc=surenb@google.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 an external index of several public inboxes,
see mirroring instructions on how to clone and mirror
all data and code used by this external index.