Linux IOMMU Development
 help / color / mirror / Atom feed
From: Heiner Kallweit <hkallweit1@gmail.com>
To: Mirsad Goran Todorovac <mirsad.todorovac@alu.unizg.hr>,
	Jason Gunthorpe <jgg@ziepe.ca>, Joerg Roedel <jroedel@suse.de>,
	Lu Baolu <baolu.lu@linux.intel.com>,
	iommu@lists.linux.dev, linux-kernel@vger.kernel.org,
	netdev@vger.kernel.org
Cc: Joerg Roedel <joro@8bytes.org>, Will Deacon <will@kernel.org>,
	Robin Murphy <robin.murphy@arm.com>,
	nic_swsd@realtek.com, "David S. Miller" <davem@davemloft.net>,
	Eric Dumazet <edumazet@google.com>,
	Jakub Kicinski <kuba@kernel.org>, Paolo Abeni <pabeni@redhat.com>,
	Marco Elver <elver@google.com>
Subject: Re: [PATCH v5 6/7] r8169: Coalesce mac ocp write and modify for 8125 and 8125B start to reduce spinlocks
Date: Mon, 30 Oct 2023 15:02:17 +0100	[thread overview]
Message-ID: <7fd8c40d-5209-4f7c-8c69-5138d0eb0bc5@gmail.com> (raw)
In-Reply-To: <20231029183600.451694-6-mirsad.todorovac@alu.unizg.hr>

On 29.10.2023 19:36, Mirsad Goran Todorovac wrote:
> Repeated calls to r8168_mac_ocp_write() and r8168_mac_ocp_modify() in
> the startup of 8125 and 8125B involve implicit spin_lock_irqsave() and
> spin_unlock_irqrestore() on each invocation.
> 
> Coalesced with the corresponding helpers r8168_mac_ocp_write_seq() and
> r8168_mac_ocp_modify_seq() into sequential write or modidy with a sinqle
> pair of spin_lock_irqsave() and spin_unlock_irqrestore(), these calls
> reduce overall lock contention.
> 
> Fixes: f1bce4ad2f1ce ("r8169: add support for RTL8125")
> Fixes: 0439297be9511 ("r8169: add support for RTL8125B")
> Cc: Heiner Kallweit <hkallweit1@gmail.com>
> Cc: Marco Elver <elver@google.com>
> Cc: nic_swsd@realtek.com
> Cc: "David S. Miller" <davem@davemloft.net>
> Cc: Eric Dumazet <edumazet@google.com>
> Cc: Jakub Kicinski <kuba@kernel.org>
> Cc: Paolo Abeni <pabeni@redhat.com>
> Cc: netdev@vger.kernel.org
> Cc: linux-kernel@vger.kernel.org
> Link: https://lore.kernel.org/lkml/20231028005153.2180411-1-mirsad.todorovac@alu.unizg.hr/
> Link: https://lore.kernel.org/lkml/20231028110459.2644926-1-mirsad.todorovac@alu.unizg.hr/
> Signed-off-by: Mirsad Goran Todorovac <mirsad.todorovac@alu.unizg.hr>
> ---
> v5:
>  added unlocked primitives to allow mac ocs modify grouping
>  applied coalescing of mac ocp writes/modifies for 8168ep and 8117
>  some formatting fixes to please checkpatch.pl
> 
> v4:
>  fixed complaints as advised by Heiner and checkpatch.pl
>  split the patch into five sections to be more easily manipulated and reviewed
>  introduced r8168_mac_ocp_write_seq()
>  applied coalescing of mac ocp writes/modifies for 8168H, 8125 and 8125B
> 
> v3:
>  removed register/mask pair array sentinels, so using ARRAY_SIZE().
>  avoided duplication of RTL_W32() call code as advised by Heiner.
> 
>  drivers/net/ethernet/realtek/r8169_main.c | 75 +++++++++++++----------
>  1 file changed, 44 insertions(+), 31 deletions(-)
> 
> diff --git a/drivers/net/ethernet/realtek/r8169_main.c b/drivers/net/ethernet/realtek/r8169_main.c
> index 50fbacb05953..0778cd0ba2e0 100644
> --- a/drivers/net/ethernet/realtek/r8169_main.c
> +++ b/drivers/net/ethernet/realtek/r8169_main.c
> @@ -3553,6 +3553,28 @@ DECLARE_RTL_COND(rtl_mac_ocp_e00e_cond)
>  
>  static void rtl_hw_start_8125_common(struct rtl8169_private *tp)
>  {
> +	static const struct e_info_regmaskset e_info_8125_common_1[] = {
> +		{ 0xd3e2, 0x0fff, 0x03a9 },
> +		{ 0xd3e4, 0x00ff, 0x0000 },
> +		{ 0xe860, 0x0000, 0x0080 },
> +	};
> +
> +	static const struct e_info_regmaskset e_info_8125_common_2[] = {
> +		{ 0xc0b4, 0x0000, 0x000c },
> +		{ 0xeb6a, 0x00ff, 0x0033 },
> +		{ 0xeb50, 0x03e0, 0x0040 },
> +		{ 0xe056, 0x00f0, 0x0030 },
> +		{ 0xe040, 0x1000, 0x0000 },
> +		{ 0xea1c, 0x0003, 0x0001 },
> +		{ 0xe0c0, 0x4f0f, 0x4403 },
> +		{ 0xe052, 0x0080, 0x0068 },
> +		{ 0xd430, 0x0fff, 0x047f },
> +		{ 0xea1c, 0x0004, 0x0000 },
> +		{ 0xeb54, 0x0000, 0x0001 },
> +	};
> +
> +	unsigned long flags;
> +
>  	rtl_pcie_state_l2l3_disable(tp);
>  
>  	RTL_W16(tp, 0x382, 0x221b);
> @@ -3560,47 +3582,38 @@ static void rtl_hw_start_8125_common(struct rtl8169_private *tp)
>  	RTL_W16(tp, 0x4800, 0);
>  
>  	/* disable UPS */
> -	r8168_mac_ocp_modify(tp, 0xd40a, 0x0010, 0x0000);
> +
> +	raw_spin_lock_irqsave(&tp->mac_ocp_lock, flags);
> +	__r8168_mac_ocp_modify(tp, 0xd40a, 0x0010, 0x0000);
>  
>  	RTL_W8(tp, Config1, RTL_R8(tp, Config1) & ~0x10);
>  
> -	r8168_mac_ocp_write(tp, 0xc140, 0xffff);
> -	r8168_mac_ocp_write(tp, 0xc142, 0xffff);
> +	__r8168_mac_ocp_write(tp, 0xc140, 0xffff);
> +	__r8168_mac_ocp_write(tp, 0xc142, 0xffff);
>  
> -	r8168_mac_ocp_modify(tp, 0xd3e2, 0x0fff, 0x03a9);
> -	r8168_mac_ocp_modify(tp, 0xd3e4, 0x00ff, 0x0000);
> -	r8168_mac_ocp_modify(tp, 0xe860, 0x0000, 0x0080);
> +	__r8168_mac_ocp_modify_seq(tp, e_info_8125_common_1);
>  
>  	/* disable new tx descriptor format */
> -	r8168_mac_ocp_modify(tp, 0xeb58, 0x0001, 0x0000);
> +	__r8168_mac_ocp_modify(tp, 0xeb58, 0x0001, 0x0000);
>  
> -	if (tp->mac_version == RTL_GIGA_MAC_VER_63)
> -		r8168_mac_ocp_modify(tp, 0xe614, 0x0700, 0x0200);
> -	else
> -		r8168_mac_ocp_modify(tp, 0xe614, 0x0700, 0x0400);
> +	if (tp->mac_version == RTL_GIGA_MAC_VER_63) {
> +		__r8168_mac_ocp_modify(tp, 0xe614, 0x0700, 0x0200);
> +		__r8168_mac_ocp_modify(tp, 0xe63e, 0x0c30, 0x0000);
> +	} else {
> +		__r8168_mac_ocp_modify(tp, 0xe614, 0x0700, 0x0400);
> +		__r8168_mac_ocp_modify(tp, 0xe63e, 0x0c30, 0x0020);
> +	}
> +
> +	__r8168_mac_ocp_modify_seq(tp, e_info_8125_common_2);
> +	raw_spin_unlock_irqrestore(&tp->mac_ocp_lock, flags);
>  
> -	if (tp->mac_version == RTL_GIGA_MAC_VER_63)
> -		r8168_mac_ocp_modify(tp, 0xe63e, 0x0c30, 0x0000);
> -	else
> -		r8168_mac_ocp_modify(tp, 0xe63e, 0x0c30, 0x0020);
> -
> -	r8168_mac_ocp_modify(tp, 0xc0b4, 0x0000, 0x000c);
> -	r8168_mac_ocp_modify(tp, 0xeb6a, 0x00ff, 0x0033);
> -	r8168_mac_ocp_modify(tp, 0xeb50, 0x03e0, 0x0040);
> -	r8168_mac_ocp_modify(tp, 0xe056, 0x00f0, 0x0030);
> -	r8168_mac_ocp_modify(tp, 0xe040, 0x1000, 0x0000);
> -	r8168_mac_ocp_modify(tp, 0xea1c, 0x0003, 0x0001);
> -	r8168_mac_ocp_modify(tp, 0xe0c0, 0x4f0f, 0x4403);
> -	r8168_mac_ocp_modify(tp, 0xe052, 0x0080, 0x0068);
> -	r8168_mac_ocp_modify(tp, 0xd430, 0x0fff, 0x047f);
> -
> -	r8168_mac_ocp_modify(tp, 0xea1c, 0x0004, 0x0000);
> -	r8168_mac_ocp_modify(tp, 0xeb54, 0x0000, 0x0001);
>  	udelay(1);
> -	r8168_mac_ocp_modify(tp, 0xeb54, 0x0001, 0x0000);
> -	RTL_W16(tp, 0x1880, RTL_R16(tp, 0x1880) & ~0x0030);
>  
> -	r8168_mac_ocp_write(tp, 0xe098, 0xc302);
> +	raw_spin_lock_irqsave(&tp->mac_ocp_lock, flags);
> +	__r8168_mac_ocp_modify(tp, 0xeb54, 0x0001, 0x0000);
> +	RTL_W16(tp, 0x1880, RTL_R16(tp, 0x1880) & ~0x0030);
> +	__r8168_mac_ocp_write(tp, 0xe098, 0xc302);
> +	raw_spin_unlock_irqrestore(&tp->mac_ocp_lock, flags);
>  
>  	rtl_loop_wait_low(tp, &rtl_mac_ocp_e00e_cond, 1000, 10);
>  

All this manual locking and unlocking makes the code harder
to read and more error-prone. Maybe, as a rule of thumb:
If you can replace a block with more than 10 mac ocp ops,
then fine with me.



  reply	other threads:[~2023-10-30 14:04 UTC|newest]

Thread overview: 16+ messages / expand[flat|nested]  mbox.gz  Atom feed  top
2023-10-29 18:35 [PATCH v5 1/7] r8169: Add r8169_mac_ocp_(write|modify)_seq helpers to reduce spinlock contention Mirsad Goran Todorovac
2023-10-29 18:35 ` [PATCH v5 2/7] r8169: Coalesce RTL8411b PHY power-down recovery calls " Mirsad Goran Todorovac
2023-10-30 13:48   ` Heiner Kallweit
2023-10-29 18:35 ` [PATCH v5 3/7] r8169: Coalesce mac ocp write and modify for 8168H start " Mirsad Goran Todorovac
2023-10-30 13:50   ` Heiner Kallweit
2023-10-29 18:36 ` [PATCH v5 4/7] r8169: Coalesce mac ocp write and modify for 8168ep " Mirsad Goran Todorovac
2023-10-30 13:51   ` Heiner Kallweit
2023-10-29 18:36 ` [PATCH v5 5/7] r8169: Reduce spinlock contention for the start of RTL8117 Mirsad Goran Todorovac
2023-10-30 13:51   ` Heiner Kallweit
2023-10-29 18:36 ` [PATCH v5 6/7] r8169: Coalesce mac ocp write and modify for 8125 and 8125B start to reduce spinlocks Mirsad Goran Todorovac
2023-10-30 14:02   ` Heiner Kallweit [this message]
2023-10-30 15:02     ` Mirsad Todorovac
2023-10-30 15:53       ` Heiner Kallweit
2023-10-29 18:36 ` [PATCH v5 7/7] r8169: Coalesce mac ocp write and modify for rtl_hw_init_8125 to reduce spinlock contention Mirsad Goran Todorovac
2023-10-30 14:03   ` Heiner Kallweit
2023-10-30 13:39 ` [PATCH v5 1/7] r8169: Add r8169_mac_ocp_(write|modify)_seq helpers " Heiner Kallweit

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=7fd8c40d-5209-4f7c-8c69-5138d0eb0bc5@gmail.com \
    --to=hkallweit1@gmail.com \
    --cc=baolu.lu@linux.intel.com \
    --cc=davem@davemloft.net \
    --cc=edumazet@google.com \
    --cc=elver@google.com \
    --cc=iommu@lists.linux.dev \
    --cc=jgg@ziepe.ca \
    --cc=joro@8bytes.org \
    --cc=jroedel@suse.de \
    --cc=kuba@kernel.org \
    --cc=linux-kernel@vger.kernel.org \
    --cc=mirsad.todorovac@alu.unizg.hr \
    --cc=netdev@vger.kernel.org \
    --cc=nic_swsd@realtek.com \
    --cc=pabeni@redhat.com \
    --cc=robin.murphy@arm.com \
    --cc=will@kernel.org \
    /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