linux-kernel - Re: [PATCH 1/1] linux/io.h: Add cleanup defination for iounmap() and memunmap()

lists.openwall.net		lists / announce owl-users owl-dev john-users john-dev passwdqc-users yescrypt popa3d-users / oss-security kernel-hardening musl sabotage tlsify passwords / crypt-dev xvendor / Bugtraq Full-Disclosure linux-kernel linux-netdev linux-ext4 linux-hardening linux-cve-announce PHC
Open Source and information security mailing list archives

Hash Suite: Windows password security audit tool. GUI, reports in PDF.

[<prev] [next>] [<thread-prev] [day] [month] [year] [list]

Message-Id: <fea17390-f3de-470b-b252-e80f3ea2135f@app.fastmail.com>
Date: Fri, 02 Aug 2024 10:07:38 +0200
From: "Arnd Bergmann" <arnd@...db.de>
To: "Frank Li" <Frank.li@....com>
Cc: "Jason Gunthorpe" <jgg@...pe.ca>,
 "Niklas Schnelle" <schnelle@...ux.ibm.com>,
 "Andrew Morton" <akpm@...ux-foundation.org>,
 "Alexei Starovoitov" <ast@...nel.org>,
 "Kent Overstreet" <kent.overstreet@...ux.dev>,
 "open list" <linux-kernel@...r.kernel.org>, imx@...ts.linux.dev
Subject: Re: [PATCH 1/1] linux/io.h: Add cleanup defination for iounmap() and
 memunmap()

On Thu, Aug 1, 2024, at 17:15, Frank Li wrote:
> On Thu, Aug 01, 2024 at 09:27:40AM +0200, Arnd Bergmann wrote:
>> On Thu, Aug 1, 2024, at 01:31, Frank Li wrote:
>> > diff --git a/include/linux/io.h b/include/linux/io.h
>> > index 59ec5eea696c4..7695d7973c8ff 100644
>> > --- a/include/linux/io.h
>> > +++ b/include/linux/io.h
>> > @@ -163,6 +163,9 @@ enum {
>> >  void *memremap(resource_size_t offset, size_t size, unsigned long flags);
>> >  void memunmap(void *addr);
>> >
>> > +DEFINE_FREE(iounmap, void __iomem *, if (!IS_ERR_OR_NULL(_T)) iounmap(_T))
>> > +DEFINE_FREE(memunmap, void *, if (!IS_ERR_OR_NULL(_T)) memunmap(_T))
>>
>> I don't like the use of IS_ERR_OR_NULL(), which tends
>> to indicate a problem in the interface design.
>
> Just !(_T) ? I just refer kfree()'s implementation.

Right, I think !(_T) is what we want here. I did see that kfree
has the same construct you are adding here, but did not read
up on how it got there.

>> In which cases do you expect to need scope based cleanup
>> on an error pointer here? The only interfaces I see that
>> returns an __iomem error pointer are the devm_* ones, but
>> those have their own cleanup method.
>
>
> devm_* can help much especial in probe() function. but scope base cleanup
> also useful.

Yes, that's fine. My only point here is that you can't combine
the cleanup function you added with devm_ioremap(), as that needs
to be paired with either implicit cleanup or with devm_iounmap() but
not plain iounmap().

    Arnd