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]
Date:   Wed, 21 Dec 2016 10:51:18 -0700
From:   Ross Zwisler <ross.zwisler@...ux.intel.com>
To:     Jan Kara <jack@...e.cz>
Cc:     Dan Williams <dan.j.williams@...el.com>, akpm@...ux-foundation.org,
        Matthew Wilcox <mawilcox@...rosoft.com>,
        Dave Chinner <david@...morbit.com>,
        linux-kernel@...r.kernel.org, Dave Hansen <dave.hansen@...el.com>,
        Alexander Viro <viro@...iv.linux.org.uk>,
        linux-fsdevel@...r.kernel.org,
        Ross Zwisler <ross.zwisler@...ux.intel.com>,
        Christoph Hellwig <hch@....de>
Subject: Re: [PATCH] dax: kill uml support

On Wed, Dec 21, 2016 at 09:53:47AM +0100, Jan Kara wrote:
> On Tue 20-12-16 17:37:40, Dan Williams wrote:
> > The lack of common transparent-huge-page helpers for UML is becoming
> > increasingly painful for fs/dax.c now that it is growing more pmd
> > functionality. Add UML to the list of unsupported architectures, and
> > clean up no-longer-necessary ifdef as a result.
> ...
> > diff --git a/fs/dax.c b/fs/dax.c
> > index ddcddfeaa03b..86df835783ea 100644
> > --- a/fs/dax.c
> > +++ b/fs/dax.c
> > @@ -710,8 +710,7 @@ static void dax_mapping_entry_mkclean(struct address_space *mapping,
> >  		if (follow_pte_pmd(vma->vm_mm, address, &ptep, &pmdp, &ptl))
> >  			continue;
> >  
> > -		if (pmdp) {
> > -#ifdef CONFIG_FS_DAX_PMD
> > +		if (pmdp && IS_ENABLED(CONFIG_FS_DAX_PMD)) {
> >  			pmd_t pmd;
> 
> So I was under the impression that pmdp can never be != NULL when
> CONFIG_FS_DAX_PMD is disabled. Otherwise Ross' patch would leave ptl locked
> in that case... Did I miss something or we can just remove IS_ENABLED()
> check?

We need the IS_ENABLED() check to prevent a different build error.

The #ifdef was there to prevent compile time errors where pmd_pfn(),
pmd_write(), etc were all undefined symbols because they aren't defined in
the arch/um headers.

For x86_64 we can get linker errors if CONFIG_TRANSPARENT_HUGEPAGE isn't set:

fs/built-in.o: In function `dax_writeback_mapping_range.part.27':
dax.c:(.text+0x4ce28): undefined reference to `pmdp_huge_clear_flush'

In this config we do have a prototype for pmdp_huge_clear_flush() in
include/asm-generic/pgtable.h so it doesn't show up as an undefined symbol,
but the implementation in mm/pgtable-generic.c is in a
#ifdef CONFIG_TRANSPARENT_HUGEPAGE block.

The IS_ENABLED() lets the compiler optimize out the code that calls
pmdp_huge_clear_flush() so it doesn't try and resolve that symbol at link
time.

(If any of this sounds incorrect or I'm using the wrong terminology for
anything, please correct me!)

You're right that 'pmdp' should always be NULL if CONFIG_FS_DAX_PMD isn't
defined.

Powered by blists - more mailing lists

Powered by Openwall GNU/*/Linux Powered by OpenVZ