Groups | Search | Server Info | Keyboard shortcuts | Login | Register [http] [https] [nntp] [nntps]

Groups > linux.kernel > #1336263 > unrolled thread

[PATCH] lightnvm: do not reserve lun on l2p loading

Back to article view | Back to linux.kernel

Contents

  [PATCH] lightnvm: do not reserve lun on l2p loading "Javier González" <jg@lightnvm.io> - 2016-02-17 12:20 +0100
    Re: [PATCH] lightnvm: do not reserve lun on l2p loading Matias Bjørling <mb@lightnvm.io> - 2016-02-18 16:40 +0100

#1336263 — [PATCH] lightnvm: do not reserve lun on l2p loading

When the l2p table is loaded, addresses are checked for the lun they
belong to and luns are reserved accordingly. This assumes that metadata
is being stored in the backend device to recover the previous target
configuration. Since this is not yet implemented, this check collides
with some of the core initialization (e.g., sysblock initialization when
a page is formed by several sectors).

We take this check out and for now rely on that the right target will be
created instead. When metadata is stored to recover a target, this check
will come natural as aprt of the recovery strategy.

FIXES c87ea75

Signed-off-by: Javier González <javier@cnexlabs.com>
---
 drivers/lightnvm/gennvm.c | 3 ---
 1 file changed, 3 deletions(-)

diff --git a/drivers/lightnvm/gennvm.c b/drivers/lightnvm/gennvm.c
index b97801c..42c1c2a 100644
--- a/drivers/lightnvm/gennvm.c
+++ b/drivers/lightnvm/gennvm.c
@@ -192,9 +192,6 @@ static int gennvm_block_map(u64 slba, u32 nlb, __le64 *entries, void *private)
 		lun_id = div_u64(pba, dev->sec_per_lun);
 		lun = &gn->luns[lun_id];
 
-		if (!test_bit(lun_id, dev->lun_map))
-			__set_bit(lun_id, dev->lun_map);
-
 		/* Calculate block offset into lun */
 		pba = pba - (dev->sec_per_lun * lun_id);
 		blk = &lun->vlun.blocks[div_u64(pba, dev->sec_per_blk)];
-- 
2.1.4

[toc] | [next] | [standalone]

#1337464

On 02/17/2016 12:16 PM, Javier González wrote:
> When the l2p table is loaded, addresses are checked for the lun they
> belong to and luns are reserved accordingly. This assumes that metadata
> is being stored in the backend device to recover the previous target
> configuration. Since this is not yet implemented, this check collides
> with some of the core initialization (e.g., sysblock initialization when
> a page is formed by several sectors).
>
> We take this check out and for now rely on that the right target will be
> created instead. When metadata is stored to recover a target, this check
> will come natural as aprt of the recovery strategy.
>
> FIXES c87ea75
>
> Signed-off-by: Javier González <javier@cnexlabs.com>
> ---
>   drivers/lightnvm/gennvm.c | 3 ---
>   1 file changed, 3 deletions(-)
>
> diff --git a/drivers/lightnvm/gennvm.c b/drivers/lightnvm/gennvm.c
> index b97801c..42c1c2a 100644
> --- a/drivers/lightnvm/gennvm.c
> +++ b/drivers/lightnvm/gennvm.c
> @@ -192,9 +192,6 @@ static int gennvm_block_map(u64 slba, u32 nlb, __le64 *entries, void *private)
>   		lun_id = div_u64(pba, dev->sec_per_lun);
>   		lun = &gn->luns[lun_id];
>
> -		if (!test_bit(lun_id, dev->lun_map))
> -			__set_bit(lun_id, dev->lun_map);
> -
>   		/* Calculate block offset into lun */
>   		pba = pba - (dev->sec_per_lun * lun_id);
>   		blk = &lun->vlun.blocks[div_u64(pba, dev->sec_per_blk)];
>

Thanks Javier, applied.

-Matias

[toc] | [prev] | [standalone]

Back to top | Article view | linux.kernel

csiph-web