Re: [Bug] mtd: rawnand: gpmi

[Miquel Raynal <miquel.raynal@bootlin.com>]

Hi Sean,

Sean Nyekjaer <sean@geanix.com> wrote on Thu, 19 Sep 2019 13:21:56
+0200:

> On 10/09/2019 13.51, Sean Nyekjaer wrote:
> > 
> > 
> > On 10/09/2019 13.08, Sascha Hauer wrote:  
> >> On Tue, Sep 10, 2019 at 01:00:30PM +0200, Sean Nyekjaer wrote:  
> >>>
> >>>
> >>> On 10/09/2019 12.48, Sascha Hauer wrote:  
> >>>> On Tue, Sep 10, 2019 at 12:18:25PM +0200, Sean Nyekjaer wrote:  
> >>>>>
> >>>>>
> >>>>> On 10/09/2019 11.55, Sascha Hauer wrote:  
> >>>>>>> [    2.434057] Bad block table written to 0x00001ffc0000, version >>>>>>> 0x01
> >>>>>>> [    2.437254] Bad block table written to 0x00001ff80000, version >>>>>>> 0x01  
> >>>>>> What about this "Bad block table written" message? You should see >>>>>> this
> >>>>>> exactly once. Do you see this multiple times, especially when >>>>>> switching
> >>>>>> kernels between the good one and the bad one?
> >>>>>>
> >>>>>> Sascha  
> >>>>>
> >>>>> Not exactly sure what you mean, but here is the dumps:
> >>>>>
> >>>>> Before (mtd: rawnand: gpmi: Implement exec_op)
> >>>>> [    3.389352] Bad block table written to 0x00001ffc0000, version 0x01
> >>>>> [    3.399019] Bad block table written to 0x00001ff80000, version 0x01
> >>>>>
> >>>>> After
> >>>>> [    3.301096] Bad block table written to 0x00001ffc0000, version 0x01
> >>>>> [    3.310599] Bad block table written to 0x00001ff80000, version 0x01  
> >>>>
> >>>> The Bad block table is written once. When you see this message multiple
> >>>> times then this means that Linux can't read the BBT and writes it >>>> again.
> >>>> So the question is: Start the good kernel multiple times. Do you see
> >>>> this message once or on each boot? Then start the bad Kernel multiple
> >>>> times. Do you see the message once or on each boot?
> >>>>
> >>>> Sascha
> >>>>  
> >>>
> >>> U-boot:  
> >>> => nand erase.chip  
> >>>
> >>> NAND erase.chip: device 0 whole chip
> >>> Skipping bad block at  0x0c780000
> >>> Skipping bad block at  0x18000000
> >>> Skipping bad block at  0x18040000
> >>> Skipping bad block at  0x1ff00000
> >>> Skipping bad block at  0x1ff40000
> >>> Skipping bad block at  0x1ff80000
> >>> Skipping bad block at  0x1ffc0000
> >>>
> >>> Look weird it marks the bbt location bad ?  
> >>
> >> Yes, that's normal. The BBT itself is marked as bad. Otherwise the they
> >> would just be used by regular mtd users.
> >>  
> >>> Or is it a uboot feature?
> >>> I have tried another board, and uboot marks the bbt location bad on >>> that as
> >>> well
> >>>
> >>> First boot:
> >>> [    4.149870] nand: device found, Manufacturer ID: 0x98, Chip ID: 0xdc
> >>>
> >>>
> >>> [    4.156589] nand: Toshiba NAND 512MiB 3,3V 8-bit
> >>> [    4.161500] nand: 512 MiB, SLC, erase size: 256 KiB, page size: >>> 4096, OOB
> >>> size: 128
> >>>
> >>> [    4.175918] Bad block table not found for chip 0
> >>> [    4.184059] Bad block table not found for chip 0
> >>> [    4.188808] Scanning device for bad blocks
> >>> [    4.690183] Bad eraseblock 798 at 0x00000c780000
> >>> [    5.155504] Bad eraseblock 1536 at 0x000018000000
> >>> [    5.161008] Bad eraseblock 1537 at 0x000018040000
> >>> [    5.487883] Bad block table written to 0x00001ffc0000, version 0x01  
> >>
> >> And is this the bad kernel or the good kernel? The question I am trying
> >> to answer is: Can the good kernel read the BBT it has written? Can the
> >> bad Kernel do that?  
> > 
> > The "First boot" and "Second boot" was before the exec_op patch...
> > 
> > This is the new kernel including the exec_op patch:
> > [    1.343615] nand: device found, Manufacturer ID: 0x98, Chip ID: 0xdc
> > 
> > [    1.343656] nand: Toshiba NAND 512MiB 3,3V 8-bit
> > 
> > [    1.343693] nand: 512 MiB, SLC, erase size: 256 KiB, page size: 4096, > OOB size: 128
> > [    1.348666] random: fast init done
> > [    1.349518] Bad block table not found for chip 0
> > [    1.351451] Bad block table not found for chip 0
> > [    1.351486] Scanning device for bad blocks
> > [    1.827337] Bad eraseblock 798 at 0x00000c780000
> > 
> > [    2.265949] Bad eraseblock 1536 at 0x000018000000
> > [    2.266318] Bad eraseblock 1537 at 0x000018040000
> > 
> > [    2.572820] Bad block table written to 0x00001ffc0000, version 0x01
> > [    2.576120] Bad block table written to 0x00001ff80000, version 0x01
> > 
> > [    2.577087] 3 fixed-partitions partitions found on MTD device gpmi-nand
> > [    2.577127] Creating 3 MTD partitions on "gpmi-nand":
> > [    2.577188] 0x000000000000-0x000000800000 : "boot"
> > 
> > 
> > 
> > [    2.584162] 0x000000800000-0x00001ca00000 : "ubi"
> > [    2.608571] 0x00001ca00000-0x000020000000 : "testing"
> > [    2.614136] gpmi-nand 1806000.gpmi-nand: driver registered.
> > 
> > Exactly the same output... which must mean it fails reading/writing the > bbt on the 4.19.x series kernel.
> > 
> > /Sean  
> 
> Hi Sascha
> 
> Please let me know when you have some time to look into this :-)
> I dosen't seem right that it writes the bbt on a 4.19 series kernel twice
> 

For me the disturbing part is:

> >>> [    4.175918] Bad block table not found for chip 0
> >>> [    4.184059] Bad block table not found for chip 0

Writing the BBT twice is expected.

Thanks,
Miquèl

______________________________________________________
Linux MTD discussion mailing list
http://lists.infradead.org/mailman/listinfo/linux-mtd/