| Message ID | 20170823002650.3133-5-nefelim4ag@gmail.com (mailing list archive) |
|---|---|
| State | New, archived |
| Headers | show |
El miércoles, 23 de agosto de 2017 2:26:48 (CEST) Timofey Titovets escribió: > + for (i = 0; i < workspace->sample_size; i += sizeof(zero)) { > + if (memcmp(&workspace->sample[i], &zero, sizeof(zero))) > + return false; Instead of just checking for 0, wouldn't it be a better idea to check for any kind of repetitions? As in, iterate over the sample and memcmp() each part of sample with the previous one. The cost would be the same, and it would detect not just zeros, but any kind of repeated data. Is there any reason I'm missing for not doing this? -- To unsubscribe from this list: send the line "unsubscribe linux-btrfs" in the body of a message to majordomo@vger.kernel.org More majordomo info at http://vger.kernel.org/majordomo-info.html
2017-08-23 20:55 GMT+03:00 Diego Calleja <diegocg@gmail.com>: > El miércoles, 23 de agosto de 2017 2:26:48 (CEST) Timofey Titovets escribió: >> + for (i = 0; i < workspace->sample_size; i += sizeof(zero)) { >> + if (memcmp(&workspace->sample[i], &zero, sizeof(zero))) >> + return false; > > Instead of just checking for 0, wouldn't it be a better idea to check > for any kind of repetitions? > > As in, iterate over the sample and memcmp() each part of sample with > the previous one. The cost would be the same, and it would detect not > just zeros, but any kind of repeated data. Is there any reason I'm > missing for not doing this? Thank you, i was not think about that, That approach seems better, i will update the patch.
diff --git a/fs/btrfs/heuristic.c b/fs/btrfs/heuristic.c index 5336638a3b7c..4557ea1db373 100644 --- a/fs/btrfs/heuristic.c +++ b/fs/btrfs/heuristic.c @@ -73,6 +73,21 @@ static struct list_head *heuristic_alloc_workspace(void) return ERR_PTR(-ENOMEM); } +static bool sample_zeroed(struct workspace *workspace) +{ + u32 i; + u8 zero[READ_SIZE]; + + memset(&zero, 0, sizeof(zero)); + + for (i = 0; i < workspace->sample_size; i += sizeof(zero)) { + if (memcmp(&workspace->sample[i], &zero, sizeof(zero))) + return false; + } + + return true; +} + static int heuristic(struct list_head *ws, struct inode *inode, u64 start, u64 end) { @@ -110,6 +125,9 @@ static int heuristic(struct list_head *ws, struct inode *inode, workspace->sample_size = b; + if (sample_zeroed(workspace)) + return 1; + memset(workspace->bucket, 0, sizeof(*workspace->bucket)*BUCKET_SIZE); for (a = 0; a < workspace->sample_size; a++) {
Use memcmp for check sample data to zeroes. Signed-off-by: Timofey Titovets <nefelim4ag@gmail.com> --- fs/btrfs/heuristic.c | 18 ++++++++++++++++++ 1 file changed, 18 insertions(+) -- 2.14.1 -- To unsubscribe from this list: send the line "unsubscribe linux-btrfs" in the body of a message to majordomo@vger.kernel.org More majordomo info at http://vger.kernel.org/majordomo-info.html