lightnvm: pblk: guarantee that backpointer is respected on writer stall

pblk's write buffer must guarantee that it respects the device's
constrains for reads (i.e., mw_cunits). This is done by maintaining a
backpointer that updates the L2P table as entries wrap up, making them
point to the media instead of pointing to the write buffer.

This mechanism can race in case that the write thread stalls, as the
write pointer will protect the last written entry, thus disregarding the
read constrains.

This patch adds an extra check on wrap up, making sure that the
threshold is respected at all times, preventing new entries to overwrite
committed data, also in case of write thread stall.

Reported-by: Heiner Litz <hlitz@ucsc.edu>
Signed-off-by: Javier González <javier@cnexlabs.com>
Reviewed-by: Heiner Litz <hlitz@ucsc.edu>
Signed-off-by: Matias Bjørling <mb@lightnvm.io>
Signed-off-by: Jens Axboe <axboe@kernel.dk>

1 files changed, 3 insertions, 2 deletions

diff --git a/drivers/lightnvm/pblk-init.c b/drivers/lightnvm/pblk-init.c
index e5239aba806b..13822594647c 100644
--- a/drivers/lightnvm/pblk-init.c
+++ b/drivers/lightnvm/pblk-init.c
@@ -193,8 +193,9 @@ static int pblk_rwb_init(struct pblk *pblk)
         struct nvm_tgt_dev *dev = pblk->dev;
         struct nvm_geo *geo = &dev->geo;
         unsigned long buffer_size;
-        int pgs_in_buffer;
+        int pgs_in_buffer, threshold;
 
+        threshold = geo->mw_cunits * geo->all_luns;
         pgs_in_buffer = (max(geo->mw_cunits, geo->ws_opt) + geo->ws_opt)
                                                                 * geo->all_luns;
 
@@ -203,7 +204,7 @@ static int pblk_rwb_init(struct pblk *pblk)
         else
                 buffer_size = pgs_in_buffer;
 
-        return pblk_rb_init(&pblk->rwb, buffer_size, geo->csecs);
+        return pblk_rb_init(&pblk->rwb, buffer_size, threshold, geo->csecs);
 }
 
 /* Minimum pages needed within a lun */