On 2/1/21 4:28 PM, Goffredo Baroncelli wrote:
From: Goffredo Baroncelli <kreij...@inwind.it>
When this mode is enabled, the chunk allocation policy is modified as follow.
Each disk may have a different tag:
- BTRFS_DEV_ALLOCATION_PREFERRED_METADATA
- BTRFS_DEV_ALLOCATION_METADATA_ONLY
- BTRFS_DEV_ALLOCATION_DATA_ONLY
- BTRFS_DEV_ALLOCATION_PREFERRED_DATA (default)
Where:
- ALLOCATION_PREFERRED_X means that it is preferred to use this disk for the
X chunk type (the other type may be allowed when the space is low)
- ALLOCATION_X_ONLY means that it is used *only* for the X chunk type. This
means also that it is a preferred choice.
Each time the allocator allocates a chunk of type X , first it takes the disks
tagged as ALLOCATION_X_ONLY or ALLOCATION_PREFERRED_X; if the space is not
enough, it uses also the disks tagged as ALLOCATION_METADATA_ONLY; if the space
is not enough, it uses also the other disks, with the exception of the one
marked as ALLOCATION_PREFERRED_Y, where Y the other type of chunk (i.e. not X).
Signed-off-by: Goffredo Baroncelli <kreij...@inwind.it>
---
fs/btrfs/volumes.c | 81 +++++++++++++++++++++++++++++++++++++++++++++-
fs/btrfs/volumes.h | 1 +
2 files changed, 81 insertions(+), 1 deletion(-)
diff --git a/fs/btrfs/volumes.c b/fs/btrfs/volumes.c
index 68b346c5465d..57ee3e2fdac0 100644
--- a/fs/btrfs/volumes.c
+++ b/fs/btrfs/volumes.c
@@ -4806,13 +4806,18 @@ static int btrfs_add_system_chunk(struct btrfs_fs_info
*fs_info,
}
/*
- * sort the devices in descending order by max_avail, total_avail
+ * sort the devices in descending order by alloc_hint,
+ * max_avail, total_avail
*/
static int btrfs_cmp_device_info(const void *a, const void *b)
{
const struct btrfs_device_info *di_a = a;
const struct btrfs_device_info *di_b = b;
+ if (di_a->alloc_hint > di_b->alloc_hint)
+ return -1;
+ if (di_a->alloc_hint < di_b->alloc_hint)
+ return 1;
if (di_a->max_avail > di_b->max_avail)
return -1;
if (di_a->max_avail < di_b->max_avail)
@@ -4939,6 +4944,15 @@ static int gather_device_info(struct btrfs_fs_devices
*fs_devices,
int ndevs = 0;
u64 max_avail;
u64 dev_offset;
+ int hint;
+
+ static const char alloc_hint_map[BTRFS_DEV_ALLOCATION_MASK_COUNT] = {
+ [BTRFS_DEV_ALLOCATION_DATA_ONLY] = -1,
+ [BTRFS_DEV_ALLOCATION_PREFERRED_DATA] = 0,
+ [BTRFS_DEV_ALLOCATION_METADATA_ONLY] = 1,
+ [BTRFS_DEV_ALLOCATION_PREFERRED_METADATA] = 2
+ /* the other values are set to 0 */
+ };
This can be made global, up with the btrfs_raid_array definitions.
/*
* in the first pass through the devices list, we gather information
@@ -4991,16 +5005,81 @@ static int gather_device_info(struct btrfs_fs_devices
*fs_devices,
devices_info[ndevs].max_avail = max_avail;
devices_info[ndevs].total_avail = total_avail;
devices_info[ndevs].dev = device;
+
+ if (((ctl->type & BTRFS_BLOCK_GROUP_DATA) &&
+ (ctl->type & BTRFS_BLOCK_GROUP_METADATA)) ||
+ info->allocation_hint_mode ==
+ BTRFS_ALLOCATION_HINT_DISABLED) {
+ /*
+ * if mixed bg or the allocator hint is
+ * disable, set all the alloc_hint
+ * fields to the same value, so the sorting
+ * is not affected
+ */
+ devices_info[ndevs].alloc_hint = 0;
+ } else if(ctl->type & BTRFS_BLOCK_GROUP_DATA) {
+ hint = device->type & BTRFS_DEV_ALLOCATION_MASK;
+
+ /*
+ * skip BTRFS_DEV_METADATA_ONLY disks
+ */
+ if (hint == BTRFS_DEV_ALLOCATION_METADATA_ONLY)
+ continue;
+ /*
+ * if a data chunk must be allocated,
+ * sort also by hint (data disk
+ * higher priority)
+ */
+ devices_info[ndevs].alloc_hint = -alloc_hint_map[hint];
+ } else { /* BTRFS_BLOCK_GROUP_METADATA */
+ hint = device->type & BTRFS_DEV_ALLOCATION_MASK;
+
+ /*
+ * skip BTRFS_DEV_DATA_ONLY disks
+ */
+ if (hint == BTRFS_DEV_ALLOCATION_DATA_ONLY)
+ continue;
+ /*
+ * if a data chunk must be allocated,
+ * sort also by hint (metadata hint
+ * higher priority)
+ */
+ devices_info[ndevs].alloc_hint = alloc_hint_map[hint];
+ }
+
++ndevs;
}
ctl->ndevs = ndevs;
+ /*
+ * no devices available
+ */
+ if (!ndevs)
+ return 0;
+
/*
* now sort the devices by hole size / available space
*/
sort(devices_info, ndevs, sizeof(struct btrfs_device_info),
btrfs_cmp_device_info, NULL);
+ /*
+ * select the minimum set of disks grouped by hint that
+ * can host the chunk
+ */
+ ndevs = 0;
+ while (ndevs < ctl->ndevs) {
+ hint = devices_info[ndevs++].alloc_hint;
+ while (devices_info[ndevs].alloc_hint == hint &&
+ ndevs < ctl->ndevs)
+ ndevs++;
+ if (ndevs >= ctl->devs_min)
+ break;
+ }
Can we just adjust btrfs_cmp_device_info to take the hint info into account?
Thanks,
Josef