Re: [flashrom] Macronix MX25L6445E

[Stefan Reinauer]

* Wagner, Helge (GE Intelligent Platforms)  [110527 15:46]:
> To be sure that the spi_block_erase_20 works for both the MX25L6405 and the
> MX25L6445E, we could change the block erase size from 64K to 4K. This should
> work even for the MX25L6405, but with the side effect of the erase taking
> longer than needed.
>
> Any comments?

works great for me!
> Please find my patches included.
>
> Signed-off-by: Helge Wagner

Acked-by: Stefan Reinauer

Macronix MX25L6445E

[Carl-Daniel Hailfinger]

Am 18.07.2011 19:07 schrieb Stefan Reinauer:
> Wagner, Helge (GE Intelligent Platforms)  [110527 15:46]:
> > To be sure that the spi_block_erase_20 works for both the MX25L6405 and the
> > MX25L6445E, we could change the block erase size from 64K to 4K. This should
> > work even for the MX25L6405, but with the side effect of the erase taking
> > longer than needed.
> >
> > Any comments?
>
> works great for me!
> > Please find my patches included.
> >
> > Signed-off-by: Helge Wagner
>
> Acked-by: Stefan Reinauer

Sorry, but AFAICS this will cause chip corruption. To be exact,
specifying a too small eraseblock size (4 kB instead of 64 kB) means
that flashrom will walk the device in 4 kB blocks, erase each block and
write to it. Now if the actual erase block size is bigger than the
specified block size, you'll get this:

Erase @0 kB: 0-4 kB assumed, 0-64 kB actual. No problem.
Write @0-4 kB: No problem.
Erase @4 kB: 4-8 kB assumed, 0-64 kB actual. You just deleted 0-4 kB again.
Write @4-8 kB: No problem.
Erase @8 kB: 8-12 kB assumed, 0-64 kB actual. You just deleted 0-8 kB again.
....
Erase @60 kB: 60-64 kB assumed, 0-64 kB actual. You just deleted 0-60 kB
again.
Write @60-64 kB: No problem.

End result: 0-60 kB blank, 60-64 kB has the desired data.

We either have to be able to distinguish between both chips at probe
time or we force the user to make a selection or we have flashrom detect
this special evil twin problem and make flashrom use only those
functions which have identical characteristics for both chips (there is
a semi-finished patch for the last variant).

Regards,
Carl-Daniel

Macronix MX25L6445E

[Stefan Reinauer]

* Carl-Daniel Hailfinger  [110718 23:44]:
> Am 18.07.2011 19:07 schrieb Stefan Reinauer:
> > Wagner, Helge (GE Intelligent Platforms)  [110527 15:46]:
> > > To be sure that the spi_block_erase_20 works for both the MX25L6405 and the
> > > MX25L6445E, we could change the block erase size from 64K to 4K. This should
> > > work even for the MX25L6405, but with the side effect of the erase taking
> > > longer than needed.
> > >
> > > Any comments?
> >
> > works great for me!
> > > Please find my patches included.
> > >
> > > Signed-off-by: Helge Wagner
> >
> > Acked-by: Stefan Reinauer
>
> Sorry, but AFAICS this will cause chip corruption. To be exact,
> specifying a too small eraseblock size (4 kB instead of 64 kB) means
> that flashrom will walk the device in 4 kB blocks, erase each block and
> write to it. Now if the actual erase block size is bigger than the
> specified block size, you'll get this:

Yes, I agree that changing the block erase size is risky. Good thing the
patch does not change it, so it should be save to apply, plus it solves
the problem.

Stefan

Macronix MX25L6445E

[Wagner, Helge (GE Germany)]

Stefan,

Did you use my latest patch (at
http://www.flashrom.org/pipermail/flashrom/2011-June/007013.html )?

Carl-Daniel,
> > > To be sure that the spi_block_erase_20 works for both the MX25L6405
> > > and the MX25L6445E, we could change the block erase size from 64K 
> > > to 4K. This should work even for the MX25L6405, but with the side 
> > > effect of the erase taking longer than needed.

This was the comment for my first try.
> Sorry, but AFAICS this will cause chip corruption. To be exact, 
> specifying a too small eraseblock size (4 kB instead of 64 kB) means 
> that flashrom will walk the device in 4 kB blocks, erase each block 
> and write to it. Now if the actual erase block size is bigger than the
> specified block size, you'll get this:

And exactly because of that I have reworked the patch (see above).

Regards,
Helge

-----Original Message-----
From: Stefan Reinauer [mailto:[e-mail]] 
Sent: Montag, 18. Juli 2011 23:49
To: Carl-Daniel Hailfinger
Cc: Wagner, Helge (GE Germany); [e-mail]
Subject: Re: [flashrom] Macronix MX25L6445E

* Carl-Daniel Hailfinger  [110718
23:44]:
> Am 18.07.2011 19:07 schrieb Stefan Reinauer:
> > Wagner, Helge (GE Intelligent Platforms)

[110527 15:46]:
> > > To be sure that the spi_block_erase_20 works for both the MX25L6405
> > > and the MX25L6445E, we could change the block erase size from 64K 
> > > to 4K. This should work even for the MX25L6405, but with the side 
> > > effect of the erase taking longer than needed.
> > >
> > > Any comments?
> >
> > works great for me!
> > > Please find my patches included.
> > >
> > > Signed-off-by: Helge Wagner
> >
> > Acked-by: Stefan Reinauer
>
> Sorry, but AFAICS this will cause chip corruption. To be exact, 
> specifying a too small eraseblock size (4 kB instead of 64 kB) means 
> that flashrom will walk the device in 4 kB blocks, erase each block 
> and write to it. Now if the actual erase block size is bigger than the
> specified block size, you'll get this:

Yes, I agree that changing the block erase size is risky. Good thing the
patch does not change it, so it should be save to apply, plus it solves
the problem.

Stefan