[<prev] [next>] [<thread-prev] [thread-next>] [day] [month] [year] [list]
Message-ID: <20110614115959.GI2353@legolas.emea.dhcp.ti.com>
Date: Tue, 14 Jun 2011 15:00:00 +0300
From: Felipe Balbi <balbi@...com>
To: Tony Lindgren <tony@...mide.com>
Cc: Janusz Krzysztofik <jkrzyszt@....icnet.pl>,
"linux-omap@...r.kernel.org" <linux-omap@...r.kernel.org>,
linux-arm-kernel@...ts.infradead.org, linux-kernel@...r.kernel.org
Subject: Re: [PATCH 3.0-rc2] OMAP: ams-delta: fix broken uevent sysfs entries
On Tue, Jun 14, 2011 at 04:48:48AM -0700, Tony Lindgren wrote:
> * Janusz Krzysztofik <jkrzyszt@....icnet.pl> [110606 18:15]:
> >
> > Removing __initdata tags, introduced by commit
> > bdc58fb950a3a1b0bc3cbd8e23d21ecdde2ac9a2, "arm: omap1: fix a bunch of
> > section mismatches", from corresponding platform_device structures
> > declared in arch/arm/mach-omap1/board-ams-delta.c, corrects the problem
> > for me, which may indicate that their members (.name ?) are still
> > referred to during runtime so they shouldn't be freed after boot.
>
> Sounds like this needs a bit more research where this initdata
> gets used?
to me it sounds like missing __init/__exit annotations on the driver.
See ams-delta-leds for instance:
static int ams_delta_led_probe(struct platform_device *pdev)
static int ams_delta_led_remove(struct platform_device *pdev)
which means those drivers will have probe sitting outside or .init.text
and trying to reference name which sits in .init.text ??? Could that be
the case here ?
But it could also be that the platform_device shouldn't be marked
__initdata.
--
balbi
Download attachment "signature.asc" of type "application/pgp-signature" (491 bytes)
Powered by blists - more mailing lists