From: Fan Li <fanofcode.li@samsung.com>
To: "'Jaegeuk Kim'" <jaegeuk@kernel.org>, "'Chao Yu'" <chao2.yu@samsung.com>
Cc: linux-fsdevel@vger.kernel.org, linux-kernel@vger.kernel.org,
linux-f2fs-devel@lists.sourceforge.net
Subject: RE: [f2fs-dev] [PATCH] f2fs: refactor __find_rev_next_{zero}_bit
Date: Fri, 06 Nov 2015 13:41:30 +0800 [thread overview]
Message-ID: <001201d11855$e863dac0$b92b9040$@samsung.com> (raw)
In-Reply-To: <20151022163623.GA51514@jaegeuk-mac02.mot.com>
In bitmaps of f2fs, bytes are sorted in ascending order, but bits in a byte are sorted
in descending order. It seems to be the reason why __reverse_ulong is needed.
May I ask why f2fs bitmap apply such order?
> -----Original Message-----
> From: Jaegeuk Kim [mailto:jaegeuk@kernel.org]
> Sent: Friday, October 23, 2015 12:36 AM
> To: Chao Yu
> Cc: linux-fsdevel@vger.kernel.org; linux-kernel@vger.kernel.org; linux-f2fs-devel@lists.sourceforge.net
> Subject: Re: [f2fs-dev] [PATCH] f2fs: refactor __find_rev_next_{zero}_bit
>
> Hi Chao,
>
> On Thu, Oct 22, 2015 at 07:24:01PM +0800, Chao Yu wrote:
> > Hi Jaegeuk,
> >
> > > -----Original Message-----
> > > From: Jaegeuk Kim [mailto:jaegeuk@kernel.org]
> > > Sent: Thursday, October 22, 2015 6:30 AM
> > > To: linux-kernel@vger.kernel.org; linux-fsdevel@vger.kernel.org;
> > > linux-f2fs-devel@lists.sourceforge.net
> > > Subject: Re: [f2fs-dev] [PATCH] f2fs: refactor
> > > __find_rev_next_{zero}_bit
> > >
> > > On Wed, Oct 21, 2015 at 02:16:43PM -0700, Jaegeuk Kim wrote:
> > > > This patch refactors __find_rev_next_{zero}_bit which was disabled
> > > > previously due to bugs.
> >
> > Actually I didn't know the history here, could you please explain
> > details about the bugs? then I can do the review.
>
> When I disabled the existing function by:
>
> commit e19ef527aa32f057710ec842fe656bffc263b0bb
> Author: Jaegeuk Kim <jaegeuk@kernel.org>
> Date: Mon May 18 11:45:15 2015 -0700
>
> f2fs: avoid buggy functions
>
> I found a mismatch of results between the existing __find_rev_next_bit and f2fs_test_bit work. I remember it happened on
> __find_rev_next_bit sometimes by xfstests which include trim_fs.
> (I didn't get an error from __find_rev_zero_bit though.)
>
> Since I have no time to investigate the bug in more detail at that time, I just disabled it. But, now I'm trying to refactor the
flow to fix
> this.
>
> Thanks,
>
> >
> > Thanks,
> >
> > > >
> > > > Signed-off-by: Jaegeuk Kim <jaegeuk@kernel.org>
> > > > ---
> > > > fs/f2fs/segment.c | 106
> > > > +++++++++++++++++++++++++-----------------------------
> > > > 1 file changed, 49 insertions(+), 57 deletions(-)
> > > >
> > > > diff --git a/fs/f2fs/segment.c b/fs/f2fs/segment.c index
> > > > f37c212..50afc93 100644
> > > > --- a/fs/f2fs/segment.c
> > > > +++ b/fs/f2fs/segment.c
> > > > @@ -29,6 +29,21 @@ static struct kmem_cache *discard_entry_slab;
> > > > static struct kmem_cache *sit_entry_set_slab; static struct
> > > > kmem_cache *inmem_entry_slab;
> > > >
> > > > +static unsigned long __reverse_ulong(unsigned char *str) {
> > > > + unsigned long tmp = 0;
> > > > + int shift = 24, idx = 0;
> > > > +
> > > > +#if BITS_PER_LONG == 64
> > > > + shift = 56;
> > > > +#endif
> > > > + while (shift >= 0) {
> > > > + tmp |= (unsigned long)str[idx++] << shift;
> > > > + shift -= BITS_PER_BYTE;
> > > > + }
> > > > + return tmp;
> > > > +}
> > > > +
> > > > /*
> > > > * __reverse_ffs is copied from include/asm-generic/bitops/__ffs.h since
> > > > * MSB and LSB are reversed in a byte by f2fs_set_bit.
> > > > @@ -38,27 +53,31 @@ static inline unsigned long __reverse_ffs(unsigned long word)
> > > > int num = 0;
> > > >
> > > > #if BITS_PER_LONG == 64
> > > > - if ((word & 0xffffffff) == 0) {
> > > > + if ((word & 0xffffffff00000000) == 0)
> > >
> > > I fixed the sparse warning by modifying:
> > > if ((word & 0xffffffff00000000UL) == 0)
> > >
> > > Thanks,
> > >
> > > > num += 32;
> > > > + else
> > > > word >>= 32;
> > > > - }
> > > > #endif
> > > > - if ((word & 0xffff) == 0) {
> > > > + if ((word & 0xffff0000) == 0)
> > > > num += 16;
> > > > + else
> > > > word >>= 16;
> > > > - }
> > > > - if ((word & 0xff) == 0) {
> > > > +
> > > > + if ((word & 0xff00) == 0)
> > > > num += 8;
> > > > + else
> > > > word >>= 8;
> > > > - }
> > > > +
> > > > if ((word & 0xf0) == 0)
> > > > num += 4;
> > > > else
> > > > word >>= 4;
> > > > +
> > > > if ((word & 0xc) == 0)
> > > > num += 2;
> > > > else
> > > > word >>= 2;
> > > > +
> > > > if ((word & 0x2) == 0)
> > > > num += 1;
> > > > return num;
> > > > @@ -68,26 +87,16 @@ static inline unsigned long __reverse_ffs(unsigned long word)
> > > > * __find_rev_next(_zero)_bit is copied from lib/find_next_bit.c because
> > > > * f2fs_set_bit makes MSB and LSB reversed in a byte.
> > > > * Example:
> > > > - * LSB <--> MSB
> > > > - * f2fs_set_bit(0, bitmap) => 0000 0001
> > > > - * f2fs_set_bit(7, bitmap) => 1000 0000
> > > > + * MSB <--> LSB
> > > > + * f2fs_set_bit(0, bitmap) => 1000 0000
> > > > + * f2fs_set_bit(7, bitmap) => 0000 0001
> > > > */
> > > > static unsigned long __find_rev_next_bit(const unsigned long *addr,
> > > > unsigned long size, unsigned long offset) {
> > > > - while (!f2fs_test_bit(offset, (unsigned char *)addr))
> > > > - offset++;
> > > > -
> > > > - if (offset > size)
> > > > - offset = size;
> > > > -
> > > > - return offset;
> > > > -#if 0
> > > > const unsigned long *p = addr + BIT_WORD(offset);
> > > > unsigned long result = offset & ~(BITS_PER_LONG - 1);
> > > > unsigned long tmp;
> > > > - unsigned long mask, submask;
> > > > - unsigned long quot, rest;
> > > >
> > > > if (offset >= size)
> > > > return size;
> > > > @@ -97,14 +106,9 @@ static unsigned long __find_rev_next_bit(const unsigned long *addr,
> > > > if (!offset)
> > > > goto aligned;
> > > >
> > > > - tmp = *(p++);
> > > > - quot = (offset >> 3) << 3;
> > > > - rest = offset & 0x7;
> > > > - mask = ~0UL << quot;
> > > > - submask = (unsigned char)(0xff << rest) >> rest;
> > > > - submask <<= quot;
> > > > - mask &= submask;
> > > > - tmp &= mask;
> > > > + tmp = __reverse_ulong((unsigned char *)p);
> > > > + tmp &= ~0UL >> offset;
> > > > +
> > > > if (size < BITS_PER_LONG)
> > > > goto found_first;
> > > > if (tmp)
> > > > @@ -112,42 +116,34 @@ static unsigned long
> > > > __find_rev_next_bit(const unsigned long *addr,
> > > >
> > > > size -= BITS_PER_LONG;
> > > > result += BITS_PER_LONG;
> > > > + p++;
> > > > aligned:
> > > > while (size & ~(BITS_PER_LONG-1)) {
> > > > - tmp = *(p++);
> > > > + tmp = __reverse_ulong((unsigned char *)p);
> > > > if (tmp)
> > > > goto found_middle;
> > > > result += BITS_PER_LONG;
> > > > size -= BITS_PER_LONG;
> > > > + p++;
> > > > }
> > > > if (!size)
> > > > return result;
> > > > - tmp = *p;
> > > > +
> > > > + tmp = __reverse_ulong((unsigned char *)p);
> > > > found_first:
> > > > - tmp &= (~0UL >> (BITS_PER_LONG - size));
> > > > - if (tmp == 0UL) /* Are any bits set? */
> > > > + tmp &= (~0UL << (BITS_PER_LONG - size));
> > > > + if (!tmp) /* Are any bits set? */
> > > > return result + size; /* Nope. */
> > > > found_middle:
> > > > return result + __reverse_ffs(tmp); -#endif }
> > > >
> > > > static unsigned long __find_rev_next_zero_bit(const unsigned long *addr,
> > > > unsigned long size, unsigned long offset) {
> > > > - while (f2fs_test_bit(offset, (unsigned char *)addr))
> > > > - offset++;
> > > > -
> > > > - if (offset > size)
> > > > - offset = size;
> > > > -
> > > > - return offset;
> > > > -#if 0
> > > > const unsigned long *p = addr + BIT_WORD(offset);
> > > > unsigned long result = offset & ~(BITS_PER_LONG - 1);
> > > > unsigned long tmp;
> > > > - unsigned long mask, submask;
> > > > - unsigned long quot, rest;
> > > >
> > > > if (offset >= size)
> > > > return size;
> > > > @@ -157,40 +153,36 @@ static unsigned long __find_rev_next_zero_bit(const unsigned long *addr,
> > > > if (!offset)
> > > > goto aligned;
> > > >
> > > > - tmp = *(p++);
> > > > - quot = (offset >> 3) << 3;
> > > > - rest = offset & 0x7;
> > > > - mask = ~(~0UL << quot);
> > > > - submask = (unsigned char)~((unsigned char)(0xff << rest) >> rest);
> > > > - submask <<= quot;
> > > > - mask += submask;
> > > > - tmp |= mask;
> > > > + tmp = __reverse_ulong((unsigned char *)p);
> > > > + tmp |= ~((~0UL << offset) >> offset);
> > > > +
> > > > if (size < BITS_PER_LONG)
> > > > goto found_first;
> > > > - if (~tmp)
> > > > + if (tmp != ~0UL)
> > > > goto found_middle;
> > > >
> > > > size -= BITS_PER_LONG;
> > > > result += BITS_PER_LONG;
> > > > + p++;
> > > > aligned:
> > > > while (size & ~(BITS_PER_LONG - 1)) {
> > > > - tmp = *(p++);
> > > > - if (~tmp)
> > > > + tmp = __reverse_ulong((unsigned char *)p);
> > > > + if (tmp != ~0UL)
> > > > goto found_middle;
> > > > result += BITS_PER_LONG;
> > > > size -= BITS_PER_LONG;
> > > > + p++;
> > > > }
> > > > if (!size)
> > > > return result;
> > > > - tmp = *p;
> > > >
> > > > + tmp = __reverse_ulong((unsigned char *)p);
> > > > found_first:
> > > > - tmp |= ~0UL << size;
> > > > - if (tmp == ~0UL) /* Are any bits zero? */
> > > > + tmp |= ~(~0UL << (BITS_PER_LONG - size));
> > > > + if (tmp == ~0UL) /* Are any bits zero? */
> > > > return result + size; /* Nope. */
> > > > found_middle:
> > > > return result + __reverse_ffz(tmp); -#endif }
> > > >
> > > > void register_inmem_page(struct inode *inode, struct page *page)
> > > > --
> > > > 2.1.1
> > >
> > > --------------------------------------------------------------------
> > > ---------- _______________________________________________
> > > Linux-f2fs-devel mailing list
> > > Linux-f2fs-devel@lists.sourceforge.net
> > > https://lists.sourceforge.net/lists/listinfo/linux-f2fs-devel
>
> ------------------------------------------------------------------------------
> _______________________________________________
> Linux-f2fs-devel mailing list
> Linux-f2fs-devel@lists.sourceforge.net
> https://lists.sourceforge.net/lists/listinfo/linux-f2fs-devel
prev parent reply other threads:[~2015-11-06 5:42 UTC|newest]
Thread overview: 5+ messages / expand[flat|nested] mbox.gz Atom feed top
2015-10-21 21:16 Jaegeuk Kim
2015-10-21 22:29 ` Jaegeuk Kim
2015-10-22 11:24 ` [f2fs-dev] " Chao Yu
2015-10-22 16:36 ` Jaegeuk Kim
2015-11-06 5:41 ` Fan Li [this message]
Reply instructions:
You may reply publicly to this message via plain-text email
using any one of the following methods:
* Save the following mbox file, import it into your mail client,
and reply-to-all from there: mbox
Avoid top-posting and favor interleaved quoting:
https://en.wikipedia.org/wiki/Posting_style#Interleaved_style
* Reply using the --to, --cc, and --in-reply-to
switches of git-send-email(1):
git send-email \
--in-reply-to='001201d11855$e863dac0$b92b9040$@samsung.com' \
--to=fanofcode.li@samsung.com \
--cc=chao2.yu@samsung.com \
--cc=jaegeuk@kernel.org \
--cc=linux-f2fs-devel@lists.sourceforge.net \
--cc=linux-fsdevel@vger.kernel.org \
--cc=linux-kernel@vger.kernel.org \
/path/to/YOUR_REPLY
https://kernel.org/pub/software/scm/git/docs/git-send-email.html
* If your mail client supports setting the In-Reply-To header
via mailto: links, try the mailto: link
Be sure your reply has a Subject: header at the top and a blank line
before the message body.
This is a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox
all inboxes | Powered by JetHome®