MIME-Version: 1.0
References: <CGME20230210045651epcas1p1392354722c8ad6d4ffad6f934ba78742@epcms1p7>
 <20230210045608.23274-1-jaewon31.kim@samsung.com> <20230210050333epcms1p7c0bd93b385828aeed9689d1d17ff6789@epcms1p7>
In-Reply-To: <20230210050333epcms1p7c0bd93b385828aeed9689d1d17ff6789@epcms1p7>
From:   John Stultz <jstultz@google.com>
Date:   Tue, 14 Feb 2023 14:25:15 -0800
Message-ID: <CANDhNCranMg0aFx_M50UtRsoNZYZ=mWSmN4_M1im19=+ZtzxiA@mail.gmail.com>
Subject: Re: [PATCH v2] dma-buf: system_heap: avoid reclaim for order 4
To:     jaewon31.kim@samsung.com
Cc:     "tjmercier@google.com" <tjmercier@google.com>,
        "sumit.semwal@linaro.org" <sumit.semwal@linaro.org>,
        "daniel.vetter@ffwll.ch" <daniel.vetter@ffwll.ch>,
        "akpm@linux-foundation.org" <akpm@linux-foundation.org>,
        "hannes@cmpxchg.org" <hannes@cmpxchg.org>,
        "mhocko@kernel.org" <mhocko@kernel.org>,
        "linux-mm@kvack.org" <linux-mm@kvack.org>,
        "linux-kernel@vger.kernel.org" <linux-kernel@vger.kernel.org>,
        "jaewon31.kim@gmail.com" <jaewon31.kim@gmail.com>
Content-Type: text/plain; charset="UTF-8"
Precedence: bulk

On Thu, Feb 9, 2023 at 9:03 PM Jaewon Kim <jaewon31.kim@samsung.com> wrote:
>
> >Using order 4 pages would be helpful for IOMMUs mapping, but trying to get order 4 pages could spend quite much time in the page allocation.
> >From the perspective of responsiveness, the deterministic memory allocation speed, I think, is quite important.
> >
> >The order 4 allocation with __GFP_RECLAIM may spend much time in reclaim and compation logic. __GFP_NORETRY also may affect. These cause unpredictable delay.
> >
> >To get reasonable allocation speed from dma-buf system heap, use HIGH_ORDER_GFP for order 4 to avoid reclaim. And let me remove meaningless __GFP_COMP for order 0.
> >
> >According to my tests, order 4 with MID_ORDER_GFP could get more number of order 4 pages but the elapsed times could be very slow.
> >
> >         time  order 8 order 4 order 0
> >     584 usec  0       160     0
> >  28,428 usec  0       160     0
> > 100,701 usec  0       160     0
> >  76,645 usec  0       160     0
> >  25,522 usec  0       160     0
> >  38,798 usec  0       160     0
> >  89,012 usec  0       160     0
> >  23,015 usec  0       160     0
> >  73,360 usec  0       160     0
> >  76,953 usec  0       160     0
> >  31,492 usec  0       160     0
> >  75,889 usec  0       160     0
> >  84,551 usec  0       160     0
> >  84,352 usec  0       160     0
> >  57,103 usec  0       160     0
> >  93,452 usec  0       160     0
> >
> >If HIGH_ORDER_GFP is used for order 4, the number of order 4 could be decreased but the elapsed time results were quite stable and fast enough.
> >
> >         time  order 8 order 4 order 0
> >   1,356 usec  0       155     80
> >   1,901 usec  0       11      2384
> >   1,912 usec  0       0       2560
> >   1,911 usec  0       0       2560
> >   1,884 usec  0       0       2560
> >   1,577 usec  0       0       2560
> >   1,366 usec  0       0       2560
> >   1,711 usec  0       0       2560
> >   1,635 usec  0       28      2112
> >     544 usec  10      0       0
> >     633 usec  2       128     0
> >     848 usec  0       160     0
> >     729 usec  0       160     0
> >   1,000 usec  0       160     0
> >   1,358 usec  0       160     0
> >   2,638 usec  0       31      2064
> >
> >Signed-off-by: Jaewon Kim <jaewon31.kim@samsung.com>
> >---
> > drivers/dma-buf/heaps/system_heap.c | 5 ++---
> > 1 file changed, 2 insertions(+), 3 deletions(-)
> >
> >diff --git a/drivers/dma-buf/heaps/system_heap.c b/drivers/dma-buf/heaps/system_heap.c
> >index e8bd10e60998..920db302a273 100644
> >--- a/drivers/dma-buf/heaps/system_heap.c
> >+++ b/drivers/dma-buf/heaps/system_heap.c
> >@@ -41,12 +41,11 @@ struct dma_heap_attachment {
> >       bool mapped;
> > };
> >
> >-#define LOW_ORDER_GFP (GFP_HIGHUSER | __GFP_ZERO | __GFP_COMP) -#define MID_ORDER_GFP (LOW_ORDER_GFP | __GFP_NOWARN)
> >+#define LOW_ORDER_GFP (GFP_HIGHUSER | __GFP_ZERO)
> > #define HIGH_ORDER_GFP  (((GFP_HIGHUSER | __GFP_ZERO | __GFP_NOWARN \
> >                               | __GFP_NORETRY) & ~__GFP_RECLAIM) \
> >                               | __GFP_COMP)
> >-static gfp_t order_flags[] = {HIGH_ORDER_GFP, MID_ORDER_GFP, LOW_ORDER_GFP};
> >+static gfp_t order_flags[] = {HIGH_ORDER_GFP, HIGH_ORDER_GFP,
> >+LOW_ORDER_GFP};
> > /*
> >  * The selection of the orders used for allocation (1MB, 64K, 4K) is designed
> >  * to match with the sizes often found in IOMMUs. Using order 4 pages instead
> >--
> >2.17.1
> >
> >
>
> added John Stultz <jstultz@google.com>

Sorry for the delay!
Reviewed-by: John Stultz <jstultz@google.com>

thanks
-john