On Wed, Mar 15, 2017 at 8:21 AM, Gerd Hoffmann [off-list ref] wrote:

  Hi,

quoted

First the flipping of the mmc host driver for the SD card slot will
cause issues for users that build it as a module. When I tested this
on Fedora the first update I ended up with a system that didn't boot.

Yep, switching drivers is a pain point here indeed.

Possibly we could fix that by adding a MODULE_SOFTDEP() to the iproc
driver?  Comments?  Better ideas?

quoted

Also i often see the device hang for a long period when systemd probes
for rfkill status, sometimes it needs to be reset but it generally
gets there.

What device is that?  rpi3 with wifi?

Yes, RPi3, with the brcmfmac43430-sdio.txt added so the firmware
loads, interface works.

quoted

[    6.529079] Hardware name: Generic DT based system
[    6.529106] [<c0311708>] (unwind_backtrace) from [<c030c438>]
(show_stack+0x18/0x1c)
[    6.529122] [<c030c438>] (show_stack) from [<c0639b44>]
(dump_stack+0x80/0xa0)
[    6.529137] [<c0639b44>] (dump_stack) from [<c034ca3c>] (__warn+0xe4/0x104)
[    6.529150] [<c034ca3c>] (__warn) from [<c034ca98>]
(warn_slowpath_fmt+0x3c/0x4c)
[    6.529166] [<c034ca98>] (warn_slowpath_fmt) from [<c0363b2c>]
(check_flush_dependency+0xac/0x134)
[    6.529184] [<c0363b2c>] (check_flush_dependency) from [<c036433c>]
(flush_work+0xa0/0x178)
[    6.529201] [<c036433c>] (flush_work) from [<c0458084>]
(drain_all_pages+0x1a8/0x1cc)
[    6.529221] [<c0458084>] (drain_all_pages) from [<c04b1d90>]
(start_isolate_page_range+0x168/0x1b4)
[    6.529239] [<c04b1d90>] (start_isolate_page_range) from
[<c045b854>] (alloc_contig_range+0xd4/0x314)
[    6.529258] [<c045b854>] (alloc_contig_range) from [<c04b6660>]
(cma_alloc+0x188/0x300)
[    6.529275] [<c04b6660>] (cma_alloc) from [<c03161e8>]
(__alloc_from_contiguous+0x40/0xd8)
[    6.529290] [<c03161e8>] (__alloc_from_contiguous) from
[<c03162bc>] (cma_allocator_alloc+0x3c/0x44)
[    6.529303] [<c03162bc>] (cma_allocator_alloc) from [<c0316504>]
(__dma_alloc+0x1d4/0x2f0)
[    6.529317] [<c0316504>] (__dma_alloc) from [<c0316698>]
(arm_dma_alloc+0x3c/0x48)
[    6.529331] [<c0316698>] (arm_dma_alloc) from [<c049d120>]
(dma_pool_alloc+0x124/0x240)
[    6.529356] [<c049d120>] (dma_pool_alloc) from [<bf009568>]
(bcm2835_dma_create_cb_chain+0xb0/0x1dc [bcm2835_dma])
[    6.529385] [<bf009568>] (bcm2835_dma_create_cb_chain
[bcm2835_dma]) from [<bf009ad4>] (bcm2835_dma_prep_slave_sg+0xf0/0x25c
[bcm2835_dma])
[    6.529418] [<bf009ad4>] (bcm2835_dma_prep_slave_sg [bcm2835_dma])
from [<bf04af04>] (bcm2835_request+0x2a4/0x3f4 [bcm2835])
[    6.529548] [<bf04af04>] (bcm2835_request [bcm2835]) from
[<bf0145d0>] (mmc_start_request+0x1f8/0x264 [mmc_core])
[    6.529756] [<bf0145d0>] (mmc_start_request [mmc_core]) from
[<bf016204>] (mmc_start_areq+0x2c8/0x318 [mmc_core])
[    6.529890] [<bf016204>] (mmc_start_areq [mmc_core]) from
[<bf07fa1c>] (mmc_blk_issue_rw_rq+0xc0/0x308 [mmc_block])
[    6.529943] [<bf07fa1c>] (mmc_blk_issue_rw_rq [mmc_block]) from
[<bf080f84>] (mmc_blk_issue_rq+0x418/0x428 [mmc_block])
[    6.529997] [<bf080f84>] (mmc_blk_issue_rq [mmc_block]) from
[<bf081138>] (mmc_queue_thread+0x148/0x1bc [mmc_block])
[    6.530031] [<bf081138>] (mmc_queue_thread [mmc_block]) from
[<c036ae30>] (kthread+0x120/0x138)
[    6.530049] [<c036ae30>] (kthread) from [<c0307e58>]
(ret_from_fork+0x14/0x3c)
[    6.530055] ---[ end trace 221a5a14ca55fa22 ]---
[    6.545765]  mmcblk0: p1 p2 p3 p4
[    6.566015] random: fast init done
[    6.623699] mmc1: new high speed SDIO card at address 0001

Looks more like a cma allocator issue on a quick glance.

Possibly, but only started once I added the dma drive into the initrd
and happens right when the MMC driver loads, we're not even past
switch root which is long before loading the vc4 driver which is the
main cma consumer.