[BUG] mtd: cfi_cmdset_0002: write regression since v4.17-rc1

[Thorsten Leemhuis]

[TLDR: adding this regression to regzbot; most of this mail is compiled
from a few templates paragraphs some of you might have seen already.]

Hi, this is your Linux kernel regression tracker speaking.

Top-posting for once, to make this easy accessible to everyone.

Thanks for the report.

Adding the regression mailing list to the list of recipients, as it
should be in the loop for all regressions, as explained here:
https://www.kernel.org/doc/html/latest/admin-guide/reporting-issues.html

To be sure this issue doesn't fall through the cracks unnoticed, I'm
adding it to regzbot, my Linux kernel regression tracking bot:

#regzbot ^introduced dfeae1073583
#regzbot title mtd: cfi_cmdset_0002: flash write accesses on the
hardware fail on a PowerPC MPC8313 to a 8-bit-parallel S29GL064N flash
#regzbot ignore-activity

Reminder: when fixing the issue, please add a 'Link:' tag with the URL
to the report (the parent of this mail), then regzbot will automatically
mark the regression as resolved once the fix lands in the appropriate
tree. For more details about regzbot see footer.

Sending this to everyone that got the initial report, to make all aware
of the tracking. I also hope that messages like this motivate people to
directly get at least the regression mailing list and ideally even
regzbot involved when dealing with regressions, as messages like this
wouldn't be needed then.

Don't worry, I'll send further messages wrt to this regression just to
the lists (with a tag in the subject so people can filter them away), as
long as they are intended just for regzbot. With a bit of luck no such
messages will be needed anyway.

Ciao, Thorsten (wearing his 'Linux kernel regression tracker' hat).

P.S.: As a Linux kernel regression tracker I'm getting a lot of reports
on my table. I can only look briefly into most of them. Unfortunately
therefore I sometimes will get things wrong or miss something important.
I hope that's not the case here; if you think it is, don't hesitate to
tell me about it in a public reply. That's in everyone's interest, as
what I wrote above might be misleading to everyone reading this; any
suggestion I gave thus might sent someone reading this down the wrong
rabbit hole, which none of us wants.

BTW, I have no personal interest in this issue, which is tracked using
regzbot, my Linux kernel regression tracking bot
(https://linux-regtracking.leemhuis.info/regzbot/). I'm only posting
this mail to get things rolling again and hence don't need to be CC on
all further activities wrt to this regression.

On 13.12.21 14:24, Ahmad Fatoum wrote:
> Hi,
> 
> I've been investigating a breakage on a PowerPC MPC8313: The SoC is connected
> via the "Enhanced Local Bus Controller" to a 8-bit-parallel S29GL064N flash,
> which is represented as a memory-mapped cfi-flash.
> 
> The regression began in v4.17-rc1 with
> 
>   dfeae1073583 ("mtd: cfi_cmdset_0002: Change write buffer to check correct value")
> 
> and causes all flash write accesses on the hardware to fail. Example output
> after v5.1-rc2[1]:
> 
>   root at host:~# mount -t jffs2 /dev/mtdblock0 /mnt
>   MTD do_write_buffer_wait(): software timeout, address:0x000c000b.
>   jffs2: Write clean marker to block at 0x000c0000 failed: -5
> 
> This issue still persists with v5.16-rc. Reverting aforementioned patch fixes
> it, but I am still looking for a change that keeps both Tokunori's and my
> hardware happy.
> 
> What Tokunori's patch did is that it strengthened the success condition
> for flash writes:
> 
>  - Prior to the patch, DQ polling was done until bits
>    stopped toggling. This was taken as an indicator that the write succeeded
>    and was reported up the stack. i.e. success condition is chip_ready()
> 
>  - After the patch, polling continues until the just written data is
>    actually read back, i.e. success condition is chip_good()
> 
> This new condition never holds for me, when DQ stabilizes, it reads 0xFF,
> never the just written data. The data is still written and can be read back
> on subsequent reads, just not at that point of time in the poll loop.
> 
> We haven't had write issues for the years predating that patch. As the
> regression has been mainline for a while, I am wondering what about my setup
> that makes it pop up here, but not elsewhere?
> 
> I consulted the data sheet[2] and found Figure 27, which describes DQ polling
> during embedded algorithms. DQ switches from status output to "True" (I assume
> True == all bits set == 0xFF) until CS# is reasserted. 
> 
> I compared with another chip's datasheet, and it (Figure 8.4) doesn't describe
> such an intermittent "True" state. In any case, the driver polls a few hundred
> times, however, before giving up, so there should be enough CS# toggles.
> 
> 
> Locally, I'll revert this patch for now. I think accepting 0xFF as a success
> condition may be appropriate, but I don't yet have the rationale to back it up.
> 
> I am investigating this some more, probably with a logic trace, but I wanted
> to report this in case someone has pointers and in case other people run into
> the same issue.
> 
> 
> Cheers,
> Ahmad
> 
> [1] Prior to d9b8a67b3b95 ("mtd: cfi: fix deadloop in cfi_cmdset_0002.c do_write_buffer") 
>     first included with v5.1-rc2, failing writes just hung indefinitely in kernel space.
>     That's fixed, but the writes still fail.
> 
> [2]: 001-98525 Rev. *B, https://www.infineon.com/dgdl/Infineon-S29GL064N_S29GL032N_64_Mbit_32_Mbit_3_V_Page_Mode_MirrorBit_Flash-DataSheet-v03_00-EN.pdf?fileId=8ac78c8c7d0d8da4017d0ed556fd548b
> 
> [3]: https://www.mouser.com/datasheet/2/268/SST39VF1601C-SST39VF1602C-16-Mbit-x16-Multi-Purpos-709008.pdf
>      Note that "true data" means valid data here, not all bits one.
>