- if_xl is locked, but not marked INTR_MPSAFE. Also, the mutex is
marked MTX_RECURSE.
- hifn is locked and INTR_MPSAFE.
- ubsec is locked and INTR_MPSAFE.
table:
- if_bge now marked as INTR_MPSAFE, SMPng locked. Busdma is now tested.
- if_dc now marked as INTR_MPSAFE, SMPng locked. Note that this driver
uses MTX_RECURSE.
- if_de no longer uses vtophys().
- if_em is now SMPng locked.
- if_en contains locking, but is not marked as INTR_MPSAFE.
- if_ep is now marked as INTR_MPSAFE, SMPng locked, but uses MTX_RECURSE.
- if_fxp is now marked SMPng locked.
- if_gx has locking, but isn't marked INTR_MPSAFE. Uses MTX_RECURSE.
- if_pcn still uses vtophys(); locking is present, but it's not marked
as INTR_MPSAFE, and the mutex is marked MTX_RECURSE.
- if_rl contains locking, but isn't marked as INTR_MPSAFE. The mutex
is marked MTX_RECURSE.
- if_sf still uses vtophys(). It contains locking, but uses MTX_RECURSE
and isn't marked INTR_MPSAFE.
- if_sis has locking, but isn't marked as INTR_MPSAFE. Its mutex is
marked MTX_RECURSE.
that I haven't yet checked for.
if_em is now INTR_MPSAFE.
Add if_ep, which is busma-safe and INTR_MPSAFE.
if_fxp is now INTR_MPSAFE.
if_sis is now INTR_MPSAFE.
Add if_wi, which is INTR_MPSAFE.
Assign sos as ata owner due to observations of practical reality.
pst is now INTR_MPSAFE.
hifn is now INTR_MPSAFE.
ubsec is now INTR_MPSAFE.
- Change `Responsible' from 'Doug White' to 'John-Mark Gurney'
as he did the busdma conversion work.
- a!=p should be clean, but not tested fully; make this known
in the `Notes' column.
- Update the `busdma' column to 'Done' as it was completed by
John-Mark quite a while ago.
Reviewed by: jmg (cursory glance)
Approved by: des (mentor)
busdma API without problems.
- It seems Doug White is working on converting the USB framework to
busdma, so assign this task to him.
- Note that bge(4) is a != p safe because the hardware has support for
64 bits addresses.
been running busdma'd fxp(4) on both platforms successfully.
Retain the note about possible problems with systems with large
amounts of memory since I don't know if this is resolved.
- Assign the firewire task to simokawa and mark it as done for
the busdma task.
safely handle sizeof(void *) != sizeof(vm_paddr_t) or !=
sizeof(bus_addr_t). In order for drivers to be supported with PAE,
they must be able to handle this, including avoiding casting of
physical addresses to pointers for printfs, pointer arithmetic,
etc.
Thought reasonable by: jake
IO addresses using kvtop(). Besides, it still uses OLDISA.
Add if_de; it has some (NetBSD-flavoured) busdma code, but uses
vtophys() wrapped in the TULIP_KVATOPHYS macro on FreeBSD.