From mboxrd@z Thu Jan 1 00:00:00 1970 Return-path: Received: from mail.kymetacorp.com ([192.81.58.21]) by bombadil.infradead.org with esmtps (Exim 4.80.1 #2 (Red Hat Linux)) id 1b7qVV-0001NP-C5 for barebox@lists.infradead.org; Tue, 31 May 2016 20:43:22 +0000 From: Trent Piepho Date: Tue, 31 May 2016 20:42:58 +0000 Message-ID: <1464727385.15779.125.camel@rtred1test09.kymeta.local> References: <1464714580-31488-1-git-send-email-andrew.smirnov@gmail.com> <1464714580-31488-9-git-send-email-andrew.smirnov@gmail.com> In-Reply-To: <1464714580-31488-9-git-send-email-andrew.smirnov@gmail.com> Content-Language: en-US Content-ID: <4A102D4CE881DE42B98A7B3723F64F12@kymetacorp.com> MIME-Version: 1.0 List-Id: List-Unsubscribe: , List-Archive: List-Post: List-Help: List-Subscribe: , Content-Type: text/plain; charset="us-ascii" Content-Transfer-Encoding: 7bit Sender: "barebox" Errors-To: barebox-bounces+u.kleine-koenig=pengutronix.de@lists.infradead.org Subject: Re: [PATCH 08/21] e1000: Consolidate register offset fixups To: Andrey Smirnov Cc: "barebox@lists.infradead.org" On Tue, 2016-05-31 at 10:09 -0700, Andrey Smirnov wrote: > Consolidate all code taking care on CSR offset differences for i210 > chips into a single place in the driver and integrate that > funcionality into E1000_{READ,WRITE}_REG macros. This way we can get > rid of all those > > if (hw->mac_type == e1000_igb) { > .... > } else { > .... > } > > snippets sprinkled all across the driver code. > > + > +static inline uint32_t e1000_true_offset(struct e1000_hw *hw, uint32_t reg) > +{ Any reason this needs to be inlined? gcc space vs size optimization choice usually does the right thing. > + if (hw->mac_type == e1000_igb) { > + unsigned int i; > + > + const struct e1000_fixup_table fixup_table[] = { > + { E1000_EEWR, E1000_I210_EEWR }, > + { E1000_PHY_CTRL, E1000_I210_PHY_CTRL }, > + { E1000_EEMNGCTL, E1000_I210_EEMNGCTL }, > + }; > + > + for (i = 0; i < ARRAY_SIZE(fixup_table); i++) { > + if (fixup_table[i].orig == reg) > + return fixup_table[i].fixed; > + } Looping through the table on each reg access seems a bit costly. What if the registers with different addresses had a flag bit in their definition? Then you could check the bit and only translate those that need translating. It would also document in the register definition that the register got translated. #define I210_ALT 0x100000 /* register has alternate addr on I210 */ #define E1000_EEWR (0x0102C | I210_ALT)/* EEPROM Write Register - RW */ if (reg & I210_ALT) { reg &= ~I210_ALT; if (hw->mac_type == e1000_igb) { /* look up alternate. note a case statement can be faster... */ case (reg) { case E1000_EEWR: reg = E1000_I210_EEWR; break; default: ; } } _______________________________________________ barebox mailing list barebox@lists.infradead.org http://lists.infradead.org/mailman/listinfo/barebox