Re: bug in mmc driver

yes it is a problem in mmc/sd controller. I will check if the no of
times we release bus matches acquiring bus.

The controller is a micron controller with this output from
# geom ls disk

Geom name: flash/spi0

Providers:

1. Name: flash/spi0

   Mediasize: 67108864 (64M)

   Sectorsize: 512

   Stripesize: 262144

   Stripeoffset: 0

   Mode: r0w0e0

   descr: s25fl512s

   ident: (null)

   rotationrate: unknown

   fwsectors: 0

   fwheads: 0



Geom name: mmcsd0

Providers:

1. Name: mmcsd0

   Mediasize: 14763950080 (14G)

   Sectorsize: 512

   Stripesize: 512

   Stripeoffset: 0

   Mode: r2w2e5

   descr: MMCHC R1J57L 1.1 SN 81687B5C MFG 11/2017 by 19 0x004e

   ident: 81687B5C

   rotationrate: 0

   fwsectors: 0

   fwheads: 0


On Sat, Oct 30, 2021 at 12:04 AM Ian Lepore <ian@freebsd.org> wrote:
>
> On Fri, 2021-10-29 at 10:02 +0530, Kamal R. Prasad wrote:
> > hello,
> >
> > if i run a stress test on a freebsd system booted from a micron mmc
> > device, and run a script which in a loop opens the same mmc device in
> > read-only mode, it causes a problem in acquiring the bus and panics
> > the box. both the stress test and a script to open the mmc device in
> > read only mode are in user space and so kernel should not have
> > panic'ed. Pl see stack trace from our freebsd derived os.
> >
> > thanks
> > -kamal
> > --------------
> >
> > panic: mmc: host bridge didn't serialize us.
> >
> > [...]
> >
>
> This indicates a problem in the low-level mmc/sd controller ("host
> bridge") driver.  You didn't say anything about what hardware is in
> use; I have no idea what driver to look at.
>
> -- Ian
>
>