Re: [PATCH] Use scsi_wait_scan on scsi devices instead of stabilized (#470628)

[Peter Jones <pjones redhat com>]

pjones wrote:
> Get rid of wait_for_scsi entirely, and make anything with a scsi
> modalias cause us to use the scsi_wait_scan kernel module.

Nevermind this for now; it doesn't work properly yet.  I'll repost when it does. 
 (Or, if hansg wants to figure out why the modalias detection around line 359 
isn't working, that'll work to...)


> ---
>  mkinitrd |   27 ++-------------------------
>  1 files changed, 2 insertions(+), 25 deletions(-)
>
> diff --git a/mkinitrd b/mkinitrd
> index 39ed66f..8c599ce 100755
> --- a/mkinitrd
> +++ b/mkinitrd
> @@ -66,7 +66,7 @@ ncryptoparts=0
>  ncryptolvs=0
>  ncryptoraids=0
>  root=""
> -wait_for_scsi="no"
> +scsi_wait_scan="no"
>  
>  NET_LIST=""
>  LD_SO_CONF=/etc/ld.so.conf
> @@ -138,7 +138,6 @@ resolve_device_name() {
>      echo nash-resolveDevice "$1" | /sbin/nash --forcequiet
>  }
>  
> -scsi_wait_scan="no"
>  findmodule() {
>      modlist=MODULES
>      subcmd="findmodule"
> @@ -359,7 +358,7 @@ findstoragedriverinsys () {
>          if [ -f modalias ]; then
>              MODALIAS=$(cat modalias)
>              if [ "${modalias::7}" == "scsi:t-" ]; then
> -                wait_for_scsi=yes
> +                scsi_wait_scan=yes
>              fi
>              moduledep $MODALIAS
>              unset MODALIAS
> @@ -1507,31 +1506,9 @@ emitmodules() {
>  EOF
>              done
>          fi
> -        # some of these aren't modules any more, but that 
> -        # generally makes them so much more likely to win
> -        # the race that we haven't seen cases where it matters
> -        # yet.  Nevertheless:
> -        # XXX PJFIX: we need a better mechanism here.
> -        if [ 1 -eq 0 \
> -                -o "ahci" == "$module" \
> -                -o "ata_piix" == "$module" \
> -                -o "ibmvscsic" == "$module" \
> -                -o "BusLogic" == "$module" \
> -                -o "mptbase" == "$module" \
> -                -o "pata_" == "${module::5}" \
> -                -o "qla" == "${module::3}" \
> -                -o "sata_" == "${module::5}" \
> -                ]; then
> -            wait_for_scsi="yes"
> -        fi
>      done
> -    if [ "$wait_for_scsi" == "yes" ]; then
> -        emit "echo Waiting for driver initialization."
> -        emit "stabilized --hash --interval 250 /proc/scsi/scsi"
> -    fi
>      unset usb_mounted
>      unset found_hcd_module
> -    unset wait_for_scsi
>  }
>  
>  if [ $ncryptodevs -ne 0 ]; then


--
        Peter

What we need is either less corruption, or more chances to
participate in it.