[<prev] [next>] [<thread-prev] [day] [month] [year] [list]
Message-ID: <20260109031204.5069-1-eleanor.lin@realtek.com>
Date: Fri, 9 Jan 2026 11:12:04 +0800
From: Yu-Chun Lin <eleanor.lin@...ltek.com>
To: <broonie@...nel.org>
CC: <baolin.wang@...ux.alibaba.com>, <cy.huang@...ltek.com>,
<cylee12@...ltek.com>, <dakr@...nel.org>, <eleanor.lin@...ltek.com>,
<gregkh@...uxfoundation.org>, <james.tai@...ltek.com>,
<linux-kernel@...r.kernel.org>, <rafael@...nel.org>,
<stanley_chang@...ltek.com>
Subject: Re: [PATCH v2] regmap: Fix race condition in hwspinlock irqsave routine
> On Wed, Jan 07, 2026 at 11:26:10AM +0800, Yu-Chun Lin wrote:
>
> > v2:
> > - Initialize 'flags' to 0. This fixes a -Werror build failure when
> > CONFIG_HWSPINLOCK is disabled, as the stub function in that case does
> > not initialize the pointer.
>
> > {
> > struct regmap *map = __map;
> > + unsigned long flags;
> >
> > hwspin_lock_timeout_irqsave(map->hwlock, UINT_MAX,
> > - &map->spinlock_flags);
> > + &flags);
> > + map->spinlock_flags = flags;
>
> I'm not seeing that initialisation in the actual patch?
Oops, sorry about that. I made a mistake and accidentally kept the v1 code.
I will fix it in v3.
Yu-Chun
Powered by blists - more mailing lists