Since commit dcebd755 ("block: use bio_for_each_bvec() to compute
multi-page bvec count"), the kernel will bug_on on the PS3 because
bio_split() is called with sectors == 0:

  kernel BUG at block/bio.c:1853!
  Oops: Exception in kernel mode, sig: 5 [#1]
  BE PAGE_SIZE=4K MMU=Hash PREEMPT SMP NR_CPUS=8 NUMA PS3
  Modules linked in: firewire_sbp2 rtc_ps3(+) soundcore ps3_gelic(+) \
  ps3rom(+) firewire_core ps3vram(+) usb_common crc_itu_t
  CPU: 0 PID: 97 Comm: blkid Not tainted 5.3.0-rc4 #1
  NIP:  c00000000027d0d0 LR: c00000000027d0b0 CTR: 0000000000000000
  REGS: c00000000135ae90 TRAP: 0700   Not tainted  (5.3.0-rc4)
  MSR:  8000000000028032 <SF,EE,IR,DR,RI>  CR: 44008240  XER: 20000000
  IRQMASK: 0
  GPR00: c000000000289368 c00000000135b120 c00000000084a500 c000000004ff8300
  GPR04: 0000000000000c00 c000000004c905e0 c000000004c905e0 000000000000ffff
  GPR08: 0000000000000000 0000000000000001 0000000000000000 000000000000ffff
  GPR12: 0000000000000000 c0000000008ef000 000000000000003e 0000000000080001
  GPR16: 0000000000000100 000000000000ffff 0000000000000000 0000000000000004
  GPR20: c00000000062fd7e 0000000000000001 000000000000ffff 0000000000000080
  GPR24: c000000000781788 c00000000135b350 0000000000000080 c000000004c905e0
  GPR28: c00000000135b348 c000000004ff8300 0000000000000000 c000000004c90000
  NIP [c00000000027d0d0] .bio_split+0x28/0xac
  LR [c00000000027d0b0] .bio_split+0x8/0xac
  Call Trace:
  [c00000000135b120] [c00000000027d130] .bio_split+0x88/0xac (unreliable)
  [c00000000135b1b0] [c000000000289368] .__blk_queue_split+0x11c/0x53c
  [c00000000135b2d0] [c00000000028f614] .blk_mq_make_request+0x80/0x7d4
  [c00000000135b3d0] [c000000000283a8c] .generic_make_request+0x118/0x294
  [c00000000135b4b0] [c000000000283d34] .submit_bio+0x12c/0x174
  [c00000000135b580] [c000000000205a44] .mpage_bio_submit+0x3c/0x4c
  [c00000000135b600] [c000000000206184] .mpage_readpages+0xa4/0x184
  [c00000000135b750] [c0000000001ff8fc] .blkdev_readpages+0x24/0x38
  [c00000000135b7c0] [c0000000001589f0] .read_pages+0x6c/0x1a8
  [c00000000135b8b0] [c000000000158c74] .__do_page_cache_readahead+0x118/0x184
  [c00000000135b9b0] [c0000000001591a8] .force_page_cache_readahead+0xe4/0xe8
  [c00000000135ba50] [c00000000014fc24] .generic_file_read_iter+0x1d8/0x830
  [c00000000135bb50] [c0000000001ffadc] .blkdev_read_iter+0x40/0x5c
  [c00000000135bbc0] [c0000000001b9e00] .new_sync_read+0x144/0x1a0
  [c00000000135bcd0] [c0000000001bc454] .vfs_read+0xa0/0x124
  [c00000000135bd70] [c0000000001bc7a4] .ksys_read+0x70/0xd8
  [c00000000135be20] [c00000000000a524] system_call+0x5c/0x70
  Instruction dump:
  7fe3fb78 482e30dc 7c0802a6 482e3085 7c9e2378 f821ff71 7ca42b78 7d3e00d0
  7c7d1b78 79290fe0 7cc53378 69290001 <0b090000> 81230028 7bca0020 7929ba62
  [ end trace 313fec760f30aa1f ]---

The problem originates from setting the segment boundary of the
request queue to -1UL. This makes get_max_segment_size() return zero
when offset is zero, whatever the max segment size. The test with
BLK_SEG_BOUNDARY_MASK fails and 'mask - (mask & offset) + 1' overflows
to zero in the return statement.

Not setting the segment boundary and using the default
value (BLK_SEG_BOUNDARY_MASK) fixes the problem.
Signed-off-by: Emmanuel Nicolet <emmanuel.nicolet@gmail.com>
Signed-off-by: Geoff Levand <geoff@infradead.org>
Signed-off-by: Michael Ellerman <mpe@ellerman.id.au>
Link: https://lore.kernel.org/r/060a416c43138f45105c0540eff1a45539f7e2fc.1589049250.git.geoff@infradead.org