Re: [PATCH 1/2] net: enc28j60: support getting irq number from gpiophandle in the device tree

[Andrew Lunn]

On Mon, Mar 10, 2025 at 10:20:01AM +0800, hanyuan-z wrote:
> Hi&nbsp;Andrew,&nbsp;<br/><br/>&gt;&nbsp;My&nbsp;understanding&nbsp;is&nbsp;that&nbsp;you&nbsp;should&nbsp;not&nbsp;need&nbsp;most&nbsp;of&nbsp;this.&nbsp;The&nbsp;IRQ<br/>&gt;&nbsp;core&nbsp;will&nbsp;handle&nbsp;converting&nbsp;a&nbsp;GPIO&nbsp;to&nbsp;an&nbsp;interrupt,&nbsp;if&nbsp;you&nbsp;just&nbsp;list<br/>&gt;&nbsp;is&nbsp;as&nbsp;an&nbsp;interrupt&nbsp;source&nbsp;in&nbsp;the&nbsp;normal&nbsp;way.<br/><br/>&gt;&nbsp;You&nbsp;say:<br/>&gt;<br/>&gt;&nbsp;&gt;&nbsp;Additionally,&nbsp;it&nbsp;is&nbsp;necessary&nbsp;for&nbsp;platforms&nbsp;that&nbsp;do&nbsp;not&nbsp;support&nbsp;pin<br/>&gt;&nbsp;&gt;&nbsp;configuration&nbsp;and&nbsp;properties&nbsp;via&nbsp;the&nbsp;device&nbsp;tree.<br/>&gt;<br/>&gt;&nbsp;Are&nbsp;you&nbsp;talking&nbsp;about&nbsp;ACPI?<br/><br/>Thanks&nbsp;for&nbsp;your&nbsp;review.&nbsp;Let&nbsp;me&nbsp;explain&nbsp;them.<br/><br/>I&nbsp;understand&nbsp;that&nbsp;specifying&nbsp;the&nbsp;interrupt&nbsp;as:<br/><br/>&nbsp;&nbsp;&nbsp;&nbsp;interrupts&nbsp;=&nbsp;&lt;&amp;gpio2&nbsp;23&nbsp;IRQ_TYPE_LEVEL_LOW&gt;;<br/><br/>would&nbsp;also&nbsp;work,&nbsp;and&nbsp;the&nbsp;IRQ&nbsp;subsystem&nbsp;will&nbsp;properly&nbsp;handle&nbsp;the<br/>conversion&nbsp;during&nbsp;the&nbsp;SPI&nbsp;probe.&nbsp;However,&nbsp;my&nbsp;problem&nbsp;is&nbsp;that&nbsp;the<br/>GPIO&nbsp;pin&nbsp;itself&nbsp;needs&nbsp;to&nbsp;be&nbsp;explicitly&nbsp;requested&nbsp;and&nbsp;configured&nbsp;a<br/>an&nbsp;input&nbsp;before&nbsp;it&nbsp;can&nbsp;be&nbsp;used&nbsp;as&nbsp;an&nbsp;IRQ&nbsp;pin.<br/><br/>My&nbsp;embedded&nbsp;platform&nbsp;has&nbsp;limited&nbsp;support,&nbsp;it&nbsp;only&nbsp;provides&nbsp;a<br/>hard-coded&nbsp;pin&nbsp;control&nbsp;driver&nbsp;and&nbsp;does&nbsp;not&nbsp;support&nbsp;configuring<br/>pinctrl&nbsp;properties&nbsp;in&nbsp;the&nbsp;device&nbsp;tree.&nbsp;So,&nbsp;there&nbsp;is&nbsp;no&nbsp;generic<br/>way&nbsp;to&nbsp;request&nbsp;the&nbsp;pin&nbsp;and&nbsp;set&nbsp;its&nbsp;direction&nbsp;via&nbsp;device&nbsp;tree&nbsp;bindings.<br/><br/>I&nbsp;noticed&nbsp;that&nbsp;some&nbsp;existing&nbsp;NIC&nbsp;drivers&nbsp;solve&nbsp;this&nbsp;issue&nbsp;by&nbsp;specifying<br/>`irq-gpios`&nbsp;in&nbsp;the&nbsp;device&nbsp;tree,&nbsp;which&nbsp;ensures&nbsp;that&nbsp;the&nbsp;GPIO&nbsp;is&nbsp;properly<br/>initialized&nbsp;before&nbsp;being&nbsp;converted&nbsp;to&nbsp;an&nbsp;IRQ.<br/><br/>That&nbsp;was&nbsp;my&nbsp;motivation&nbsp;for&nbsp;these&nbsp;patches.<br/><br/>And&nbsp;my&nbsp;changes&nbsp;are&nbsp;not&nbsp;related&nbsp;to&nbsp;ACPI&nbsp;in&nbsp;any&nbsp;way—they&nbsp;are&nbsp;mainly<br/>focused&nbsp;on&nbsp;device&nbsp;tree&nbsp;handling.<br/><br/>Thanks,&nbsp;&nbsp;<br/>Hanyuan&nbsp;Zhao&nbsp;&nbsp;<br/>

Please use plain text for emails.

	Andrew