| Message ID | 20190322144843.9602-13-igor.j.konopko@intel.com (mailing list archive) |
|---|---|
| State | New, archived |
| Headers | show |
| Series | lightnvm: next set of improvements for 5.2 | expand |
> On 22 Mar 2019, at 22.48, Igor Konopko <igor.j.konopko@intel.com> wrote: > > Emeta does not have corresponding OOB metadata mapping valid, so there > is no need to try to recover L2P mapping from it. > > Signed-off-by: Igor Konopko <igor.j.konopko@intel.com> > --- > drivers/lightnvm/pblk-recovery.c | 10 ++++++++++ > 1 file changed, 10 insertions(+) > > diff --git a/drivers/lightnvm/pblk-recovery.c b/drivers/lightnvm/pblk-recovery.c > index 4e8f382..74e5b17 100644 > --- a/drivers/lightnvm/pblk-recovery.c > +++ b/drivers/lightnvm/pblk-recovery.c > @@ -479,6 +479,15 @@ static int pblk_recov_scan_oob(struct pblk *pblk, struct pblk_line *line, > goto retry_rq; > } > > + if (paddr >= line->emeta_ssec) { > + /* > + * We reach emeta region and we don't want > + * to recover oob from emeta region. > + */ > + bio_put(bio); > + goto completed; > + } > + > This recovery path is supposed to be a last resource to recover an open line. Since emeta_ssec is calculated based on the new state after recovery, it is not guaranteed that the bad blocks are the same as when the last pblk instance was interrupted. Thus, emeta_ssec is not to be trusted. I would rather do the full scan at this point. Javier
On 25.03.2019 07:23, Javier González wrote: >> On 22 Mar 2019, at 22.48, Igor Konopko <igor.j.konopko@intel.com> wrote: >> >> Emeta does not have corresponding OOB metadata mapping valid, so there >> is no need to try to recover L2P mapping from it. >> >> Signed-off-by: Igor Konopko <igor.j.konopko@intel.com> >> --- >> drivers/lightnvm/pblk-recovery.c | 10 ++++++++++ >> 1 file changed, 10 insertions(+) >> >> diff --git a/drivers/lightnvm/pblk-recovery.c b/drivers/lightnvm/pblk-recovery.c >> index 4e8f382..74e5b17 100644 >> --- a/drivers/lightnvm/pblk-recovery.c >> +++ b/drivers/lightnvm/pblk-recovery.c >> @@ -479,6 +479,15 @@ static int pblk_recov_scan_oob(struct pblk *pblk, struct pblk_line *line, >> goto retry_rq; >> } >> >> + if (paddr >= line->emeta_ssec) { >> + /* >> + * We reach emeta region and we don't want >> + * to recover oob from emeta region. >> + */ >> + bio_put(bio); >> + goto completed; >> + } >> + >> > > This recovery path is supposed to be a last resource to recover an open > line. Since emeta_ssec is calculated based on the new state after > recovery, it is not guaranteed that the bad blocks are the same as when > the last pblk instance was interrupted. Thus, emeta_ssec is not to be > trusted. > > I would rather do the full scan at this point. > Ok, make sense, so we can drop this patch. > Javier >
diff --git a/drivers/lightnvm/pblk-recovery.c b/drivers/lightnvm/pblk-recovery.c index 4e8f382..74e5b17 100644 --- a/drivers/lightnvm/pblk-recovery.c +++ b/drivers/lightnvm/pblk-recovery.c @@ -479,6 +479,15 @@ static int pblk_recov_scan_oob(struct pblk *pblk, struct pblk_line *line, goto retry_rq; } + if (paddr >= line->emeta_ssec) { + /* + * We reach emeta region and we don't want + * to recover oob from emeta region. + */ + bio_put(bio); + goto completed; + } + pblk_get_packed_meta(pblk, rqd); bio_put(bio); @@ -499,6 +508,7 @@ static int pblk_recov_scan_oob(struct pblk *pblk, struct pblk_line *line, if (left_ppas > 0) goto next_rq; +completed: #ifdef CONFIG_NVM_PBLK_DEBUG WARN_ON(padded && !pblk_line_is_full(line)); #endif
Emeta does not have corresponding OOB metadata mapping valid, so there is no need to try to recover L2P mapping from it. Signed-off-by: Igor Konopko <igor.j.konopko@intel.com> --- drivers/lightnvm/pblk-recovery.c | 10 ++++++++++ 1 file changed, 10 insertions(+)