mmc_da: implement d_dump method, sddadump

[avg]

Summary

sddadump has been derived from sddastart.

mmc_sim interface has grown a new method, cam_poll, to support polled
operation.

mmc_sim code has been changed to provide a sim_poll hook only if the
controller implements the new method. The hooks is implemented in terms
of the new mmc_sim_cam_poll method.
Additionally, in-progress CCB-s now have CAM_REQ_INPROG status to
satisfy xpt_pollwait().

mmc_sim_cam_poll method has been implemented in dwmmc host controller.

Test Plan

Tested on Rock64:

panic: kdb_sysctl_panic
cpuid = 3
time = 1641841341
KDB: stack backtrace:
db_trace_self() at db_trace_self
db_trace_self_wrapper() at db_trace_self_wrapper+0x30
vpanic() at vpanic+0x174
panic() at panic+0x44
kdb_sysctl_panic() at kdb_sysctl_panic+0x68
sysctl_root_handler_locked() at sysctl_root_handler_locked+0xe0
sysctl_root() at sysctl_root+0x1ac
userland_sysctl() at userland_sysctl+0x140
sys___sysctl() at sys___sysctl+0x68
do_el0_sync() at do_el0_sync+0x4b0
handle_el0_sync() at handle_el0_sync+0x90
--- exception, esr 0x56000000
Uptime: 1m55s
Dumping 161 out of 996 MB:..3%..13%..23%..33%..43%..53%..62%..72%..82%..92%
Dump complete

After the reboot:

Jan 10 19:04:53 rock64b savecore[896]: reboot after panic: kdb_sysctl_panic
Writing crash summary to /var/crash/core.txt.3.

The vmcore file was usable with kgdb.
zpool scrub did not detect any damage (writes outside of the swap / dump partition).

[imp]

This seems to be sane, but I've not had time to study it in detail.

sys/cam/mmc/mmc_da.c

2040

when would that be possible during a system core dump?

[avg]

sys/cam/mmc/mmc_da.c

2040

I guess it can happen if a different eMMC partition is accessed right before a crash / panic.
But so far I haven't seen any practical use of those.

[mav]

Makes sense, just couple minors points.

sys/cam/mmc/mmc_sim.c

53

How can this be NULL if it is assigned in cam_sim_alloc() and never changed?

170–171

CAM_REQ_INPROG should already be set by xpt_action(). It should not be needed here.

[avg]

sys/cam/mmc/mmc_sim.c

53

Hmm, I just thoughtlessly copied the check from mmc_cam_sim_default_action().
I guess it can be safely removed from both places.

[avg]

sys/cam/mmc/mmc_sim.c

170–171

I see. Thank you for the advice.

[avg]

address comments from mav

[manu]

sys/cam/mmc/mmc_da.c

2055

/ MMC_SECTOR_SIZE would be more clear.

2057

We really should make a define for this but that's out of scope of this review.

2063

MMC_SECTOR_SIZE * count would be more clear

[avg]

sys/cam/mmc/mmc_da.c

2055

Let me do this as a follow-up commit (I promise!) and cover sddastart as well.
Thanks!

[manu]

sys/cam/mmc/mmc_da.c

2055

Sure :)