| Message ID | 1463184168-3381-3-git-send-email-jonathan.derrick@intel.com (mailing list archive) |
|---|---|
| State | New, archived |
| Headers | show |
> -----Original Message----- > From: linux-block-owner@vger.kernel.org [mailto:linux-block- > owner@vger.kernel.org] On Behalf Of Jon Derrick > Sent: Friday, May 13, 2016 7:03 PM ... > Subject: [RFCv2 2/3] block: add helper for setting and clearing S_DAX on > inode > > inode->i_flags locking rules suggest using the i_mutex lock. This patch > adds a helper to do the locking and setting of S_DAX on the block device > inode. ... > +static void bd_set_dax(struct block_device *bdev, bool enabled) > +{ > + struct inode *inode = bdev->bd_inode; > + > + inode_lock(inode); > + if (enabled) > + inode->i_flags = S_DAX; > + else > + inode->i_flags &= ~S_DAX; > + inode_unlock(inode); > +} This is not symmetric - setting wipes out any other bits, but clearing only clears the S_DAX bit. That seems confusing for a helper function. Using |= would be symmetric, but wouldn't replace what __blkdev_get does (if what it does is appropriate). > @@ -1206,9 +1218,9 @@ static int __blkdev_get(struct block_device *bdev, > fmode_t mode, int for_part) > bdev->bd_queue = disk->queue; > bdev->bd_contains = bdev; > if (IS_ENABLED(CONFIG_BLK_DEV_DAX) && disk->fops- > >direct_access) > - bdev->bd_inode->i_flags = S_DAX; > + bd_set_dax(bdev, 1); > else > - bdev->bd_inode->i_flags &= ~S_DAX; > + bd_set_dax(bdev, 0); -- To unsubscribe from this list: send the line "unsubscribe linux-block" in the body of a message to majordomo@vger.kernel.org More majordomo info at http://vger.kernel.org/majordomo-info.html
On Sat, May 14, 2016 at 05:05:59AM +0000, Elliott, Robert (Persistent Memory) wrote: > > > > -----Original Message----- > > From: linux-block-owner@vger.kernel.org [mailto:linux-block- > > owner@vger.kernel.org] On Behalf Of Jon Derrick > > Sent: Friday, May 13, 2016 7:03 PM > ... > > Subject: [RFCv2 2/3] block: add helper for setting and clearing S_DAX on > > inode > > > > inode->i_flags locking rules suggest using the i_mutex lock. This patch > > adds a helper to do the locking and setting of S_DAX on the block device > > inode. > ... > > +static void bd_set_dax(struct block_device *bdev, bool enabled) > > +{ > > + struct inode *inode = bdev->bd_inode; > > + > > + inode_lock(inode); > > + if (enabled) > > + inode->i_flags = S_DAX; > > + else > > + inode->i_flags &= ~S_DAX; > > + inode_unlock(inode); > > +} > > This is not symmetric - setting wipes out any other bits, but > clearing only clears the S_DAX bit. That seems confusing for > a helper function. > > Using |= would be symmetric, but wouldn't replace what > __blkdev_get does (if what it does is appropriate). > I'm under the impression that S_DAX is the only inode flag appropriate for a block device inode with direct-access, even though it does break the 'flag' semantic. It does bring up an interesting question about the fate of the block device inode when S_DAX is set and then cleared later due to the two other conditions that can clear S_DAX. I think the appropriate fix is to save the inode flags when setting DAX and restore them when we need to clear S_DAX. > > @@ -1206,9 +1218,9 @@ static int __blkdev_get(struct block_device *bdev, > > fmode_t mode, int for_part) > > bdev->bd_queue = disk->queue; > > bdev->bd_contains = bdev; > > if (IS_ENABLED(CONFIG_BLK_DEV_DAX) && disk->fops- > > >direct_access) > > - bdev->bd_inode->i_flags = S_DAX; > > + bd_set_dax(bdev, 1); > > else > > - bdev->bd_inode->i_flags &= ~S_DAX; > > + bd_set_dax(bdev, 0); > > -- > To unsubscribe from this list: send the line "unsubscribe linux-block" in > the body of a message to majordomo@vger.kernel.org > More majordomo info at http://vger.kernel.org/majordomo-info.html -- To unsubscribe from this list: send the line "unsubscribe linux-block" in the body of a message to majordomo@vger.kernel.org More majordomo info at http://vger.kernel.org/majordomo-info.html
diff --git a/fs/block_dev.c b/fs/block_dev.c index d4fa725..252e459 100644 --- a/fs/block_dev.c +++ b/fs/block_dev.c @@ -1159,6 +1159,18 @@ void bd_set_size(struct block_device *bdev, loff_t size) } EXPORT_SYMBOL(bd_set_size); +static void bd_set_dax(struct block_device *bdev, bool enabled) +{ + struct inode *inode = bdev->bd_inode; + + inode_lock(inode); + if (enabled) + inode->i_flags = S_DAX; + else + inode->i_flags &= ~S_DAX; + inode_unlock(inode); +} + static void __blkdev_put(struct block_device *bdev, fmode_t mode, int for_part); /* @@ -1206,9 +1218,9 @@ static int __blkdev_get(struct block_device *bdev, fmode_t mode, int for_part) bdev->bd_queue = disk->queue; bdev->bd_contains = bdev; if (IS_ENABLED(CONFIG_BLK_DEV_DAX) && disk->fops->direct_access) - bdev->bd_inode->i_flags = S_DAX; + bd_set_dax(bdev, 1); else - bdev->bd_inode->i_flags &= ~S_DAX; + bd_set_dax(bdev, 0); if (!partno) { ret = -ENXIO; @@ -1239,7 +1251,7 @@ static int __blkdev_get(struct block_device *bdev, fmode_t mode, int for_part) if (!ret) { bd_set_size(bdev,(loff_t)get_capacity(disk)<<9); if (!blkdev_dax_capable(bdev)) - bdev->bd_inode->i_flags &= ~S_DAX; + bd_set_dax(bdev, 0); } /* @@ -1276,7 +1288,7 @@ static int __blkdev_get(struct block_device *bdev, fmode_t mode, int for_part) } bd_set_size(bdev, (loff_t)bdev->bd_part->nr_sects << 9); if (!blkdev_dax_capable(bdev)) - bdev->bd_inode->i_flags &= ~S_DAX; + bd_set_dax(bdev, 0); } } else { if (bdev->bd_contains == bdev) {
inode->i_flags locking rules suggest using the i_mutex lock. This patch adds a helper to do the locking and setting of S_DAX on the block device inode. Signed-off-by: Jon Derrick <jonathan.derrick@intel.com> --- fs/block_dev.c | 20 ++++++++++++++++---- 1 file changed, 16 insertions(+), 4 deletions(-)