lkml.org 
[lkml]   [2009]   [Jun]   [29]   [last100]   RSS Feed
Views: [wrap][no wrap]   [headers]  [forward] 
 
Messages in this thread
    /
    Date
    From
    SubjectRe: [PATCH] ide-cd: Improve "weird block size" error message

    On Tuesday 2009-06-23 12:59, David Miller wrote:
    >> In that case I'd like to propose the following patch. Currently the error
    >> can get printed much to frequently when there's a disc in the drive.
    >> Example:
    >>
    >> Jun 13 18:06:28 gimli kernel: ide-cd: hdd: weird block size 2352
    >> Jun 13 18:06:28 gimli kernel: ide-cd: hdd: default to 2kb block size
    >> Jun 13 18:06:32 gimli kernel: ide-cd: hdd: weird block size 2352
    >> Jun 13 18:06:42 gimli kernel: ide-cd: hdd: default to 2kb block size
    >
    >Thinking about this a bit. Let's look at what problem this is
    >trying to avoid, as per the commit message:
    >
    >--------------------
    > ide-cd: fix oops when using growisofs
    >
    > cdrom_read_capacity() will blindly return the capacity from the device
    > without sanity-checking it. This later causes code in fs/buffer.c to
    > oops.
    >
    > Fix this by checking that the device is telling us sensible things.
    >--------------------
    >
    >Well, for the values Frans's CDROM is giving, this OOPS would not
    >take place and the weird sector value is completely harmless.
    >
    >Since SECTOR_BITS is 9:
    >
    >(2352 >> 9) == (2048 >> 9) == 4
    >
    >There is simply no benefit from this warning in this situation.


    But 2352 is the block size for CDDA (yes, I reckon that is not quite
    relevant for the block layer), so it's not like there would be
    garbage bits in the lower 9 bits.

    >
    >Therefore, any objections to something like this?
    >
    >ide-cd: Don't warn on bogus block size unless it actually matters.
    >
    >Frans Pop reported that his CDROM drive reports a blocksize of 2352,
    >and this causes new warnings due to commit
    >e8e7b9eb11c34ee18bde8b7011af41938d1ad667 ("ide-cd: fix oops when using
    >growisofs").
    >
    >What we're trying to do is make sure that "blocklen >> SECTOR_BITS"
    >is something the block layer won't choke on.
    >
    >And for Frans case "2352 >> SECTOR_BITS" is equal to
    >"2048 >> SECTOR_BITS", and thats "4".


    \
     
     \ /
      Last update: 2009-06-29 13:23    [W:0.026 / U:0.604 seconds]
    ©2003-2016 Jasper Spaans. hosted at Digital OceanAdvertise on this site