mmc: core: Don't cancel detect work during MMC suspend

When MMC host is suspending the cancellation of the MMC detect
work is in progress. And mmc_rescan() which had triggered for
SD card is blocked since mmcqd/1 has acquired the host lock.
The mmcqd/1 is waiting to proceed for the card status check.
Meanwhile, a write request is received on eMMC but it cannot
proceed as it is waiting for the ongoing runtime suspend to finish.

"kworker/0:1"
----
-000|need_resched()
-001|schedule_timeout(timeout = 2147483647)
-002|wait_for_common(x = 0xD9391E90, ?, state = 2, iowait = 0)
-003|flush_workqueue(wq = 0xD92DCA80)
-004|mmc_suspend_host(host = 0xD8B20000)
-005|msmsdcc_runtime_suspend(dev = 0xC08B0598)
-006|pm_generic_runtime_suspend(?)
-007|__rpm_callback(cb = 0xC02F1BD8, dev = 0xC08B0598)
-008|rpm_suspend(dev = 0xC08B0598, rpmflags = 2)
-009|pm_runtime_work(work = 0xC08B065C)
-010|process_one_work(worker = 0xD9367F40, work = 0xC08B065C)
-011|worker_thread(__worker = 0xD9367F40)
-012|kthread(_create = 0xD924FF18)
-013|kernel_thread_exit(asm)
 ---|end of frame
    |

"kworker/u:4"
----
-000|need_resched()
-001|__mmc_claim_host(host = 0xD8B21000, abort = 0x0)
-002|mmc_sd_detect(host = 0xD8B21000)
-003|mmc_rescan(work = 0xD8B212A0)
-004|process_one_work(worker = 0xCDB12440, work = 0xD8B212A0)
-005|worker_thread(__worker = 0xCDB12440)
-006|kthread(_create = 0xD8BC7F18)
-007|kernel_thread_exit(asm)
 ---|end of frame
    |

mmcqd/1 (running)
----
-000|need_resched()
-001|schedule_timeout(timeout = 2147483647)
-002|io_schedule_timeout(timeout = 2147483647)
-003|wait_for_common(x = 0xD8633E2C, ?, state = 2, iowait = 1)
-004|mmc_wait_for_req_done(host = 0xD8B21000, mrq = 0xD8633E1C)
-005|mmc_wait_for_cmd(host = 0xD8B21000, cmd = 0xD8633E50, retries = 5)
-006|get_card_status(card = 0xD8D8B800, status = 0xD8633EA4, retries = 5)
-007|mmc_blk_err_check(card = 0xD8D8B800, areq = 0xD8624440)
-008|mmc_start_req(host = 0xD8B21000, areq = 0x0, error = 0xD8633F4C)
-009|mmc_blk_issue_rw_rq(mq = 0xD8624008, ?)
-010|mmc_blk_issue_rq(mq = 0xD8624008, req = 0x0)
-011|mmc_queue_thread(d = 0xD8624008)
-012|kthread(_create = 0xD9397D80)
-013|kernel_thread_exit(asm)
 ---|end of frame
    |

mmcqd/0
----
-000|need_resched()
-001|rpm_resume(dev = 0xC08B0598, rpmflags = 4)
-002|__pm_runtime_resume(dev = 0xC08B0598, rpmflags = 4)
-003|msmsdcc_enable(mmc = 0xD8B20000)
-004|__mmc_claim_host(host = 0xD8B20000, abort = 0x0)
-005|mmc_blk_issue_rq(mq = 0xD8C43008, req = 0xC52C9798)
-006|mmc_queue_thread(d = 0xD8C43008)
-007|kthread(_create = 0xD9397D80)
-008|kernel_thread_exit(asm)
 ---|end of frame
    |

So remove the cancellation of the detect work for the
MMC suspend to proceed and avoid this deadlock.

CRs-Fixed: 471053
Change-Id: Id79d6b81e992b08039d639b1c0f6c2eb4a79df62
Signed-off-by: Pratibhasagar V <pratibha@codeaurora.org>