less aggressive bad block marking patch (was RE: [Yaffs] Re: bit error rates)

[Charles Manning]

On Saturday 11 February 2006 12:39, Michael Schmidt wrote:
> I would like to submit that the patch below might be appropriate for those
> who aren't paranoid about ECC errors on read-back contributing to a greater
> chance of future data loss in that block.
>
> Charles, is there anything I've missed?

Hi Steve

I will look into this a bit more.

The selction symbol should be CONFIG_YAFFS_xxxx.

I think a "three strikes and you're out" policy would be a GoodThing.

>
> --MichaeL
>
> --- yaffs_tagscompat-old.c 2006-02-10 16:37:01.000000000 -0700
> +++ yaffs_tagscompat.c 2006-02-10 16:37:04.000000000 -0700
> @@ -335,6 +335,7 @@
>
>  static void yaffs_HandleReadDataError(yaffs_Device * dev, int chunkInNAND)
>  {
> +#ifdef PARANOID_ABOUT_READ_ERRORS
>   int blockInNAND = chunkInNAND / dev->nChunksPerBlock;
>
>   /* Mark the block for retirement */
> @@ -347,6 +348,7 @@
>    * then retire the block
>    * NB recursion
>    */
> +#endif
>  }
>
>  #ifdef NOTYET