linux-kernel - Re: [PATCH 1/2] debugfs: improve DEFINE_DEBUGFS_ATTRIBUTE for !CONFIG_DEBUGFS

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

Message-ID: <87vawwmuw5.fsf@gmail.com>
Date:   Thu, 13 Oct 2016 11:59:54 +0200
From:   Nicolai Stange <nicstange@...il.com>
To:     Arnd Bergmann <arnd@...db.de>
Cc:     Greg Kroah-Hartman <gregkh@...uxfoundation.org>,
        Andy Shevchenko <andriy.shevchenko@...ux.intel.com>,
        Rajneesh Bhardwaj <rajneesh.bhardwaj@...el.com>,
        Darren Hart <dvhart@...ux.intel.com>,
        Nicolai Stange <nicstange@...il.com>,
        linux-kernel@...r.kernel.org
Subject: Re: [PATCH 1/2] debugfs: improve DEFINE_DEBUGFS_ATTRIBUTE for !CONFIG_DEBUGFS_FS

Hi Arnd,

thanks for this (and sorry for the late reply)!

Arnd Bergmann <arnd@...db.de> writes:

> The slp_s0_residency_usec debugfs file currently uses
> DEFINE_DEBUGFS_ATTRIBUTE(), but that macro cannot really be used to
> define files outside of the debugfs code, as it has no reference to
> the get/set functions if CONFIG_DEBUGFS_FS is not defined:
>
> drivers/platform/x86/intel_pmc_core.c:80:12: error: ‘pmc_core_dev_state_get’ defined but not used [-Werror=unused-function]
>
> This fixes the macro to always contain the reference, and instead rely
> on the stubbed-out debugfs_create_file to not actually refer to
> its arguments so the compiler can still drop the reference.
> This works because the attribute definition is always 'static',
> and the dead-code removal silently drops all static symbols
> that are not used.
>
> Fixes: c64688081490 ("debugfs: add support for self-protecting attribute file fops")
> Fixes: df2294fb6428 ("intel_pmc_core: Convert to DEFINE_DEBUGFS_ATTRIBUTE")
> Signed-off-by: Arnd Bergmann <arnd@...db.de>
> ---
>  include/linux/debugfs.h | 43 ++++++++++++++++++++-----------------------
>  1 file changed, 20 insertions(+), 23 deletions(-)
>
> diff --git a/include/linux/debugfs.h b/include/linux/debugfs.h
> index 4d3f0d1aec73..e94f5f8dced3 100644
> --- a/include/linux/debugfs.h
> +++ b/include/linux/debugfs.h
> @@ -62,6 +62,26 @@ static inline const struct file_operations *debugfs_real_fops(struct file *filp)
>  	return filp->f_path.dentry->d_fsdata;
>  }
>  
> +ssize_t debugfs_attr_read(struct file *file, char __user *buf,
> +			size_t len, loff_t *ppos);
> +ssize_t debugfs_attr_write(struct file *file, const char __user *buf,
> +			size_t len, loff_t *ppos);
> +
> +#define DEFINE_DEBUGFS_ATTRIBUTE(__fops, __get, __set, __fmt)		\
> +static int __fops ## _open(struct inode *inode, struct file *file)	\
> +{									\
> +	__simple_attr_check_format(__fmt, 0ull);			\
> +	return simple_attr_open(inode, file, __get, __set, __fmt);	\
> +}									\
> +static const struct file_operations __fops = {				\
> +	.owner	 = THIS_MODULE,						\
> +	.open	 = __fops ## _open,					\
> +	.release = simple_attr_release,					\
> +	.read	 = debugfs_attr_read,					\
> +	.write	 = debugfs_attr_write,					\

This depends on GCC dead code elimination to always work for this
situation, otherwise we'd get undefined references to
debugfs_attr_read/write(), right?

In order to avoid having to test your patch against all those older
versions of GCC, can we have a safety net here and define some dummy
debugfs_attr_read/write() for the !CONFIG_DEBUGFS case?

If nothing else, it would IMHO make the !CONFIG_DEBUGFS case more
understandable because one had not to figure out that this actually
relies on dead code elimination to work.

Thanks,

Nicolai