Skip to content

L
linux
Switch branch/tag
  1. 07 Dec, 2020 15 commits
  3. 02 Dec, 2020 9 commits
  5. 01 Dec, 2020 16 commits
    • James Smart's avatar
      scsi: lpfc: Correct null ndlp reference on routine exit · 9d8de441
      James Smart authored 
      smatch correctly called out a logic error with accessing a pointer after
checking it for null:

 drivers/scsi/lpfc/lpfc_els.c:2043 lpfc_cmpl_els_plogi()
 error: we previously assumed 'ndlp' could be null (see line 1942)

Adjust the exit point to avoid the trace printf ndlp reference. A trace
entry was already generated when the ndlp was checked for null.

Link: https://lore.kernel.org/r/20201130181226.16675-1-james.smart@broadcom.com
Fixes: 4430f7fd ("scsi: lpfc: Rework locations of ndlp reference taking")
Reported-by: default avatarDan Carpenter <dan.carpenter@oracle.com>
Signed-off-by: default avatarJames Smart <james.smart@broadcom.com>
Signed-off-by: default avatarMartin K. Petersen <martin.petersen@oracle.com>
      9d8de441
    • Can Guo's avatar
      scsi: ufs: Stop hardcoding the scale down gear · 29b87e92
      Can Guo authored 
      Instead of hardcoding the scale down gear, make it a member of
the ufs_clk_scaling struct.

Link: https://lore.kernel.org/r/1606442334-22641-1-git-send-email-cang@codeaurora.orgReviewed-by: default avatarStanley Chu <stanley.chu@mediatek.com>
Reviewed-by: default avatarBean Huo <beanhuo@micron.com>
Reviewed-by: default avatarAsutosh Das <asutoshd@codeaurora.org>
Signed-off-by: default avatarCan Guo <cang@codeaurora.org>
Signed-off-by: default avatarMartin K. Petersen <martin.petersen@oracle.com>
      29b87e92
    • Can Guo's avatar
      scsi: ufs-qcom: Keep core_clk_unipro on while link is active · 96f08cc5
      Can Guo authored 
      If we want to disable clocks to save power but still keep the link active,
core_clk_unipro, like ref_clk, should not be the one being disabled.

Link: https://lore.kernel.org/r/1606356063-38380-3-git-send-email-cang@codeaurora.orgReviewed-by: default avatarHongwu Su <hongwus@codeaurora.org>
Reviewed-by: default avatarAsutosh Das <asutoshd@codeaurora.org>
Signed-off-by: default avatarCan Guo <cang@codeaurora.org>
Signed-off-by: default avatarMartin K. Petersen <martin.petersen@oracle.com>
      96f08cc5
    • Can Guo's avatar
      scsi: ufs: Refactor ufshcd_setup_clocks() to remove skip_ref_clk · 81309c24
      Can Guo authored 
      Remove the param skip_ref_clk from __ufshcd_setup_clocks(), but keep a flag
in struct ufs_clk_info to tell whether a clock can be disabled or not while
the link is active.

Link: https://lore.kernel.org/r/1606356063-38380-2-git-send-email-cang@codeaurora.orgReviewed-by: default avatarHongwu Su <hongwus@codeaurora.org>
Reviewed-by: default avatarBean Huo <beanhuo@micron.com>
Reviewed-by: default avatarStanley Chu <stanley.chu@mediatek.com>
Signed-off-by: default avatarCan Guo <cang@codeaurora.org>
Signed-off-by: default avatarMartin K. Petersen <martin.petersen@oracle.com>
      81309c24
    • Sebastian Andrzej Siewior's avatar
      scsi: message: fusion: Remove in_interrupt() usage in mptsas_cleanup_fw_event_q() · 817a7c99
      Sebastian Andrzej Siewior authored 
      mptsas_cleanup_fw_event_q() uses in_interrupt() to determine if it is safe
to cancel a worker item.

Aside of that in_interrupt() is deprecated as it does not provide what the
name suggests. It covers more than hard/soft interrupt servicing context
and is semantically ill defined.

Looking closer there are a few problems with the current construct:

 - It could be invoked from an interrupt handler / non-blocking context
   because cancel_delayed_work() has no such restriction. Also,
   mptsas_free_fw_event() has no such restriction.

 - The list is accessed unlocked. It may dequeue a valid work-item but at
   the time of invoking cancel_delayed_work() the memory may be released or
   reused because the worker has already run.

mptsas_cleanup_fw_event_q() is invoked via mptsas_shutdown() which is
always invoked from preemtible context on device shutdown.  It is also
invoked via mptsas_ioc_reset(, MPT_IOC_POST_RESET) which is a
MptResetHandlers callback. The only caller here are mpt_SoftResetHandler(),
mpt_HardResetHandler() and mpt_Soft_Hard_ResetHandler(). All these
functions have a `sleepFlag' argument and each caller uses caller uses
`CAN_SLEEP' here and according to current documentation: | @sleepFlag:
Indicates if sleep or schedule must be called

So it is safe to sleep.

Add mptsas_hotplug_event::users member. Initialize it to one by default so
mptsas_free_fw_event() will free the memory.  mptsas_cleanup_fw_event_q()
will increment its value for items it dequeues and then it may keep a
pointer after dropping the lock.  Invoke cancel_delayed_work_sync() to
cancel the work item and wait if the worker is currently busy. Free the
memory afterwards since it owns the last reference to it.

Link: https://lore.kernel.org/r/20201126132952.2287996-15-bigeasy@linutronix.de
Cc: Sathya Prakash <sathya.prakash@broadcom.com>
Cc: Sreekanth Reddy <sreekanth.reddy@broadcom.com>
Cc: Suganath Prabu Subramani <suganath-prabu.subramani@broadcom.com>
Cc: MPT-FusionLinux.pdl@broadcom.com
Reviewed-by: default avatarDaniel Wagner <dwagner@suse.de>
Signed-off-by: default avatarSebastian Andrzej Siewior <bigeasy@linutronix.de>
Signed-off-by: default avatarMartin K. Petersen <martin.petersen@oracle.com>
      817a7c99
    • Thomas Gleixner's avatar
      scsi: message: fusion: Remove in_interrupt() usage in mpt_config() · b8a51443
      Thomas Gleixner authored 
      in_interrupt() is referenced all over the place in these drivers. Most of
these references are comments which are outdated and wrong.

Aside of that in_interrupt() is deprecated as it does not provide what the
name suggests. It covers more than hard/soft interrupt servicing context
and is semantically ill defined.

>From reading the mpt_config() code and the history this is clearly a debug
mechanism and should probably be replaced by might_sleep() or completely
removed because such checks are already in the subsequent functions.

Remove the in_interrupt() references and replace the usage in mpt_config()
with might_sleep().

Link: https://lore.kernel.org/r/20201126132952.2287996-14-bigeasy@linutronix.de
Cc: Sathya Prakash <sathya.prakash@broadcom.com>
Cc: Sreekanth Reddy <sreekanth.reddy@broadcom.com>
Cc: Suganath Prabu Subramani <suganath-prabu.subramani@broadcom.com>
Cc: MPT-FusionLinux.pdl@broadcom.com
Reviewed-by: default avatarDaniel Wagner <dwagner@suse.de>
Signed-off-by: default avatarThomas Gleixner <tglx@linutronix.de>
Signed-off-by: default avatarSebastian Andrzej Siewior <bigeasy@linutronix.de>
Signed-off-by: default avatarMartin K. Petersen <martin.petersen@oracle.com>
      b8a51443
    • Ahmed S. Darwish's avatar
      scsi: myrs: Remove WARN_ON(in_interrupt()) · ca685369
      Ahmed S. Darwish authored 
      The in_interrupt() macro is ill-defined and does not provide what the name
suggests. The usage especially in driver code is deprecated and a tree-wide
effort to clean up and consolidate the (ab)usage of in_interrupt() and
related checks is happening.

In this case the check covers only parts of the contexts in which these
functions cannot be called. It fails to detect preemption or interrupt
disabled invocations.

As wait_for_completion() already contains a broad variety of checks (always
enabled or debug option dependent) which cover all invalid conditions
already, there is no point in having extra inconsistent warnings in
drivers.

Just remove it.

Link: https://lore.kernel.org/r/20201126132952.2287996-12-bigeasy@linutronix.de
Cc: Hannes Reinecke <hare@kernel.org>
Reviewed-by: default avatarDaniel Wagner <dwagner@suse.de>
Signed-off-by: default avatarAhmed S. Darwish <a.darwish@linutronix.de>
Signed-off-by: default avatarSebastian Andrzej Siewior <bigeasy@linutronix.de>
Signed-off-by: default avatarMartin K. Petersen <martin.petersen@oracle.com>
      ca685369
    • Ahmed S. Darwish's avatar
      scsi: myrb: Remove WARN_ON(in_interrupt()) · 3bc08b95
      Ahmed S. Darwish authored 
      The in_interrupt() macro is ill-defined and does not provide what the name
suggests. The usage especially in driver code is deprecated and a tree-wide
effort to clean up and consolidate the (ab)usage of in_interrupt() and
related checks is happening.

In this case the check covers only parts of the contexts in which these
functions cannot be called. It fails to detect preemption or interrupt
disabled invocations.

As wait_for_completion() already contains a broad variety of checks (always
enabled or debug option dependent) which cover all invalid conditions
already, there is no point in having extra inconsistent warnings in
drivers.

Just remove it.

Link: https://lore.kernel.org/r/20201126132952.2287996-11-bigeasy@linutronix.de
Cc: Hannes Reinecke <hare@kernel.org>
Reviewed-by: default avatarDaniel Wagner <dwagner@suse.de>
Signed-off-by: default avatarAhmed S. Darwish <a.darwish@linutronix.de>
Signed-off-by: default avatarSebastian Andrzej Siewior <bigeasy@linutronix.de>
Signed-off-by: default avatarMartin K. Petersen <martin.petersen@oracle.com>
      3bc08b95
    • Ahmed S. Darwish's avatar
      scsi: mpt3sas: Remove in_interrupt() · 547c0d1a
      Ahmed S. Darwish authored 
      _scsih_fw_event_cleanup_queue() waits for all outstanding firmware events
wokrqueue handlers to finish. If in_interrupt() is true, it cancels itself
and return early.

That in_interrupt() check is ill-defined and does not provide what the name
suggests: it does not cover all states in which it is safe to block and
call functions like cancel_work_sync().

That check is also not needed: _scsih_fw_event_cleanup_queue() is always
invoked from process context. Below is an analysis of its callers:

  - scsih_remove(), bound to PCI ->remove(), process context

  - scsih_shutdown(), bound to PCI ->shutdown(), process context

  - mpt3sas_scsih_clear_outstanding_scsi_tm_commands(), called by
      => _base_clear_outstanding_commands(), called by
        =>_base_fault_reset_work(), workqueue
        => mpt3sas_base_hard_reset_handler(), locks mutex

Remove the in_interrupt() check. Change _scsih_fw_event_cleanup_queue()
specification to a purely process-context function and mark it with
"Context: task, can sleep".

Link: https://lore.kernel.org/r/20201126132952.2287996-10-bigeasy@linutronix.de
Cc: Sathya Prakash <sathya.prakash@broadcom.com>
Cc: Sreekanth Reddy <sreekanth.reddy@broadcom.com>
Cc: Suganath Prabu Subramani <suganath-prabu.subramani@broadcom.com>
Cc: <MPT-FusionLinux.pdl@broadcom.com>
Reviewed-by: default avatarDaniel Wagner <dwagner@suse.de>
Signed-off-by: default avatarAhmed S. Darwish <a.darwish@linutronix.de>
Signed-off-by: default avatarSebastian Andrzej Siewior <bigeasy@linutronix.de>
Signed-off-by: default avatarMartin K. Petersen <martin.petersen@oracle.com>
      547c0d1a
    • Ahmed S. Darwish's avatar
      scsi: qla4xxx: Remove in_interrupt() from qla4_82xx_rom_lock() · 014aced1
      Ahmed S. Darwish authored 
      qla4_82xx_rom_lock() spins on a certain hardware state until it is
updated. At the end of each spin, if in_interrupt() is true, it does 20
loops of cpu_relax(). Otherwise, it yields the CPU.

While in_interrupt() is ill-defined and does not provide what the name
suggests, it is not needed here: qla4_82xx_rom_lock() is always called
from process context. Below is an analysis of its callers:

  - ql4_nx.c: qla4_82xx_rom_fast_read(), all process context callers:
    => ql4_nx.c: qla4_82xx_pinit_from_rom(), GFP_KERNEL allocation
    => ql4_nx.c: qla4_82xx_load_from_flash(), msleep() in a loop

  - ql4_nx.c: qla4_82xx_pinit_from_rom(), earlier discussed

  - ql4_nx.c: qla4_82xx_rom_lock_recovery(), bound to "isp_operations"
    ->rom_lock_recovery() hook, which has one process context caller,
    qla4_8xxx_device_bootstrap(), with callers:
      => ql4_83xx.c: qla4_83xx_need_reset_handler(), process, msleep()
      => ql4_nx.c: qla4_8xxx_device_state_handler(), multiple msleep()s

  - ql4_nx.c: qla4_82xx_read_flash_data(), has cond_resched()

Remove the in_interrupt() check. Mark, qla4_82xx_rom_lock(), and the
->rom_lock_recovery() hook, with "Context: task, can sleep".

Change qla4_82xx_rom_lock() implementation to sleep 20ms, instead of a
schedule(), for each spin. This is more deterministic, and it matches
the other implementations bound to ->rom_lock_recovery().

Link: https://lore.kernel.org/r/20201126132952.2287996-9-bigeasy@linutronix.de
Cc: Nilesh Javali <njavali@marvell.com>
Cc: Manish Rangankar <mrangankar@marvell.com>
Cc: <GR-QLogic-Storage-Upstream@marvell.com>
Reviewed-by: default avatarDaniel Wagner <dwagner@suse.de>
Signed-off-by: default avatarAhmed S. Darwish <a.darwish@linutronix.de>
Signed-off-by: default avatarSebastian Andrzej Siewior <bigeasy@linutronix.de>
Signed-off-by: default avatarMartin K. Petersen <martin.petersen@oracle.com>
      014aced1
    • Ahmed S. Darwish's avatar
      scsi: qla4xxx: Remove in_interrupt() from qla4_82xx_idc_lock() · 3627668c
      Ahmed S. Darwish authored 
      qla4_82xx_idc_lock() spins on a certain hardware state until it is
updated. At the end of each spin, if in_interrupt() is true, it does 20
loops of cpu_relax(). Otherwise, it yields the CPU.

While in_interrupt() is ill-defined and does not provide what the name
suggests, it is not needed here: qla4_82xx_idc_lock() is always called from
process context. Below is an analysis of its callers:

  - ql4_nx.c: qla4_82xx_need_reset_handler(), 1-second msleep() in a
    loop.

  - ql4_nx.c: qla4_82xx_isp_reset(), calls
    qla4_8xxx_device_state_handler(), which has multiple msleep()s.

Beside direct calls, qla4_82xx_idc_lock() is also bound to isp_operations
->idc_lock() hook. Other functions which are bound to the same hook,
e.g. qla4_83xx_drv_lock(), also have an msleep(). For completeness, below
is an analysis of all callers of that hook:

  - ql4_83xx.c: qla4_83xx_need_reset_handler(), has an msleep()

  - ql4_83xx.c: qla4_83xx_isp_reset(), calls
    qla4_8xxx_device_state_handler(), which has multiple msleep()s.

  - ql4_83xx.c: qla4_83xx_disable_pause(), all process context callers:
    => ql4_mbx.c: qla4xxx_mailbox_command(), msleep(), mutex_lock()
    => ql4_os.c: qla4xxx_recover_adapter(), schedule_timeout() in loop
    => ql4_os.c: qla4xxx_do_dpc(), workqueue context

  - ql4_attr.c: qla4_8xxx_sysfs_write_fw_dump(), sysfs bin_attribute
    ->write() hook, process context

  - ql4_mbx.c: qla4xxx_mailbox_command(), earlier discussed

  - ql4_nx.c: qla4_8xxx_device_bootstrap(), callers:
    => ql4_83xx.c: qla4_83xx_need_reset_handler(), process, msleep()
    => ql4_nx.c: qla4_8xxx_device_state_handler(), earlier discussed

  - ql4_nx.c: qla4_8xxx_need_qsnt_handler(), callers:
    => ql4_nx.c: qla4_8xxx_device_state_handler(), multiple msleep()s
    => ql4_os.c: qla4xxx_do_dpc(), workqueue context

  - ql4_nx.c: qla4_8xxx_update_idc_reg(), callers:
    => ql4_nx.c: qla4_8xxx_device_state_handler(), earlier discussed
    => ql4_os.c: qla4_8xxx_error_recovery(), only called by
    qla4xxx_pci_slot_reset(), which is bound to PCI ->slot_reset()
    process-context hook

  - ql4_nx.c: qla4_8xxx_device_state_handler(), earlier discussed

  - ql4_os.c: qla4xxx_recover_adapter(), earlier discussed

  - ql4_os.c: qla4xxx_do_dpc(), earlier discussed

Remove the in_interrupt() check. Mark, qla4_82xx_idc_lock(), and the
->idc_lock() hook itself, with "Context: task, can sleep".

Change qla4_82xx_idc_lock() implementation to sleep 100ms, instead of a
schedule(), for each spin. This is more deterministic, and it matches other
PCI HW locking functions in the driver.

Link: https://lore.kernel.org/r/20201126132952.2287996-8-bigeasy@linutronix.de
Cc: Nilesh Javali <njavali@marvell.com>
Cc: Manish Rangankar <mrangankar@marvell.com>
Cc: <GR-QLogic-Storage-Upstream@marvell.com>
Reviewed-by: default avatarDaniel Wagner <dwagner@suse.de>
Signed-off-by: default avatarAhmed S. Darwish <a.darwish@linutronix.de>
Signed-off-by: default avatarSebastian Andrzej Siewior <bigeasy@linutronix.de>
Signed-off-by: default avatarMartin K. Petersen <martin.petersen@oracle.com>
      3627668c
    • Ahmed S. Darwish's avatar
      scsi: qla2xxx: Remove in_interrupt() from qla83xx-specific code · 4f6a57c2
      Ahmed S. Darwish authored 
      qla83xx_wait_logic() is used to control the frequency of device IDC lock
retries. If in_interrupt() is true, it does 20 loops of cpu_relax().
Otherwise, it sleeps for 100ms and yields the CPU.

While in_interrupt() is ill-defined and does not provide what the name
suggests, it is not needed here: that qla83xx_wait_logic() is exclusively
called by qla83xx_idc_lock() / unlock(), and they always run from process
context. Below is an analysis of all the idc lock/unlock callers, in order
of appearance:

  - qla_os.c:
      qla83xx_nic_core_unrecoverable_work(),
      qla83xx_idc_state_handler_work(),
      qla83xx_nic_core_reset_work(),
      qla83xx_service_idc_aen(), all workqueue context

  - qla_os.c: qla83xx_check_nic_core_fw_alive(), has msleep()

  - qla_os.c: qla83xx_set_drv_presence(), called once from
    qla2x00_abort_isp(), which is bound to process-context ->abort_isp()
    hook. It also invokes wait_for_completion_timeout() through the chain
    qla2x00_configure_hba() => qla24xx_link_initialize() =>
    qla2x00_mailbox_command().

  - qla_os.c: qla83xx_clear_drv_presence(), which is called from
    qla2x00_abort_isp() discussed above, and from qla2x00_remove_one()
    which is PCI process-context ->remove() hook.

  - qla_os.c: qla83xx_need_reset_handler(), has a one second msleep() in
    a loop.

  - qla_os.c: qla83xx_device_bootstrap(), called only by
    qla83xx_idc_state_handler(), which has multiple msleep()
    invocations.

  - qla_os.c: qla83xx_idc_state_handler(), multiple msleep()
    invocations.

  - qla_attr.c: qla2x00_sysfs_write_reset(), sysfs bin_attribute
    ->write() hook, process context

  - qla_init.c: qla83xx_nic_core_fw_load()
      => qla_init.c: qla2x00_initialize_adapter()
        => bound to isp_operations ->initialize_adapter() hook
        ** => qla_os.c: qla2x00_probe_one(), PCI ->probe() process ctx

  - qla_init.c: qla83xx_initiating_reset(), msleep() in a loop.

  - qla_init.c: qla83xx_nic_core_reset(), called by
    qla83xx_nic_core_reset_work(), workqueue context.

Remove the in_interrupt() check, and thus replace the entirety of
qla83xx_wait_logic() with an msleep(QLA83XX_WAIT_LOGIC_MS).

Mark qla83xx_idc_lock() / unlock() with "Context: task, can sleep".

Link: https://lore.kernel.org/r/20201126132952.2287996-7-bigeasy@linutronix.de
Cc: Nilesh Javali <njavali@marvell.com>
Cc: GR-QLogic-Storage-Upstream@marvell.com
Reviewed-by: default avatarHimanshu Madhani <himanshu.madhani@oracle.com>
Reviewed-by: default avatarDaniel Wagner <dwagner@suse.de>
Signed-off-by: default avatarAhmed S. Darwish <a.darwish@linutronix.de>
Signed-off-by: default avatarSebastian Andrzej Siewior <bigeasy@linutronix.de>
Signed-off-by: default avatarMartin K. Petersen <martin.petersen@oracle.com>
      4f6a57c2
    • Ahmed S. Darwish's avatar
      scsi: target: tcm_qla2xxx: Remove BUG_ON(in_interrupt()) · 9fef41f2
      Ahmed S. Darwish authored 
      tcm_qla2xxx_free_session() has a BUG_ON(in_interrupt()).

While in_interrupt() is ill-defined and does not provide what the name
suggests, it is not needed here: the function is always invoked from
workqueue context through "struct qla_tgt_func_tmpl" ->free_session() hook
it is bound to.

The function also calls wait_event_timeout() down the chain, which already
has a might_sleep().

Remove the in_interrupt() check.

Link: https://lore.kernel.org/r/20201126132952.2287996-6-bigeasy@linutronix.de
Cc: Nilesh Javali <njavali@marvell.com>
Cc: <GR-QLogic-Storage-Upstream@marvell.com>
Reviewed-by: default avatarHimanshu Madhani <himanshu.madhani@oracle.com>
Reviewed-by: default avatarDaniel Wagner <dwagner@suse.de>
Signed-off-by: default avatarAhmed S. Darwish <a.darwish@linutronix.de>
Signed-off-by: default avatarSebastian Andrzej Siewior <bigeasy@linutronix.de>
Signed-off-by: default avatarMartin K. Petersen <martin.petersen@oracle.com>
      9fef41f2
    • Ahmed S. Darwish's avatar
      scsi: qla2xxx: Remove in_interrupt() from qla82xx-specific code · 8ac246bd
      Ahmed S. Darwish authored 
      qla82xx_idc_lock() spins on a certain hardware state until it's updated. At
the end of each spin, if in_interrupt() is true, it does 20 loops of
cpu_relax(). Otherwise, it yields the CPU.

While in_interrupt() is ill-defined and does not provide what the name
suggests, it is not needed here: qla82xx_idc_lock() is always called from
process context. Below is an analysis of its callers, in order of
appearance:

  - qla_nx.c: qla82xx_device_bootstrap(), only called by
    qla82xx_device_state_handler(), has multiple msleep()s.

  - qla_nx.c: qla82xx_need_qsnt_handler(), has one second msleep()

  - qla_nx.c: qla82xx_wait_for_state_change(), one second msleep()

  - qla_nx.c: qla82xx_need_reset_handler(), can sleep up to 10 seconds

  - qla_nx.c: qla82xx_device_state_handler(), has multiple msleep()s

  - qla_nx.c: qla82xx_abort_isp(), if it's a qla82xx controller, calls
    qla82xx_device_state_handler(), which sleeps. It's also bound to
    isp_operations ->abort_isp() hook, where all the callers are in process
    context.

  - qla_nx.c: qla82xx_beacon_on(), bound to isp_operations ->beacon_on()
    hook.  That hook is only called once, in a mutex locked context, from
    qla2x00_beacon_store().

  - qla_nx.c: qla82xx_beacon_off(), bound to isp_operations ->beacon_off()
    hook.  Like ->beacon_on(), it's only called once, in a mutex locked
    context, from qla2x00_beacon_store().

  - qla_nx.c: qla82xx_fw_dump(), calls qla2x00_wait_for_chip_reset(), which
    has msleep() in a loop. It is bound to isp_operations ->fw_dump()
    hook. That hook *is* called from atomic context at qla_isr.c by
    multiple interrupt handlers. Nonetheless, it's other controllers
    interrupt handlers, and not the qla82xx.

  - qla82xx_msix_default() and qla82xx_msix_rsp_q() call
    qla24xx_process_response_queue() which doesn't implement the firmware
    dumping.

  - qla_attr.c: qla2x00_sysfs_write_fw_dump(), and
    qla2x00_sysfs_write_reset(), process-context sysfs ->write() hooks.

  - qla_os.c: qla2x00_probe_one(). PCI ->probe(), process context.

  - qla_os.c: qla2x00_clear_drv_active(), called solely from
    qla2x00_remove_one(), which is PCI ->remove() hook, process context.

  - qla_os.c: qla2x00_do_dpc(), kthread function, process context.

Remove the in_interrupt() check. Change qla82xx_idc_lock() specification to
a purely process-context function. Mark it with "Context: task, might
sleep".

Change qla82xx_idc_lock() implementation to sleep 100ms, instead of a
schedule(), for each spin. This is more deterministic, and it matches the
other qla models idc_lock() functions.

Link: https://lore.kernel.org/r/20201126132952.2287996-5-bigeasy@linutronix.de
Cc: Nilesh Javali <njavali@marvell.com>
Cc: <GR-QLogic-Storage-Upstream@marvell.com>
Reviewed-by: default avatarHimanshu Madhani <himanshu.madhani@oracle.com>
Reviewed-by: default avatarDaniel Wagner <dwagner@suse.de>
Signed-off-by: default avatarAhmed S. Darwish <a.darwish@linutronix.de>
Signed-off-by: default avatarSebastian Andrzej Siewior <bigeasy@linutronix.de>
Signed-off-by: default avatarMartin K. Petersen <martin.petersen@oracle.com>
      8ac246bd
    • Ahmed S. Darwish's avatar
      scsi: qla4xxx: Remove in_interrupt() · a93c3835
      Ahmed S. Darwish authored 
      qla4_82xx_crb_win_lock() spins on a certain hardware state until it's
updated. At the end of each spin, if in_interrupt() is true, it does 20
loops of cpu_relax(). Otherwise, it yields the CPU.

The in_interrupt() macro is ill-defined as it does not provide what the
name suggests, and it does not catch the intended use-case here.

qla4_82xx_crb_win_lock() is always invoked with scsi_qla_host::hw_lock
acquired, with disabled interrupts. If the caller is in process context, as
in qla4_82xx_need_reset_handler(), then in_interrupt() will return false
even though it is not allowed to call schedule().

Remove the in_interrupt() check.

Change qla4_82xx_crb_win_lock() specification to a purely atomic
function. Mark it as static, remove its forward declaration, and move it
above its callers. To avoid hammering the PCI bus while spinning, use a 10
micro-second delay instead of cpu_relax().

Link: https://lore.kernel.org/r/20201126132952.2287996-4-bigeasy@linutronix.de
Fixes: f4f5df23 ("[SCSI] qla4xxx: Added support for ISP82XX")
Cc: Nilesh Javali <njavali@marvell.com>
Cc: Manish Rangankar <mrangankar@marvell.com>
Cc: <GR-QLogic-Storage-Upstream@marvell.com>
Reviewed-by: default avatarDaniel Wagner <dwagner@suse.de>
Signed-off-by: default avatarAhmed S. Darwish <a.darwish@linutronix.de>
Signed-off-by: default avatarSebastian Andrzej Siewior <bigeasy@linutronix.de>
Signed-off-by: default avatarMartin K. Petersen <martin.petersen@oracle.com>
      a93c3835
    • Ahmed S. Darwish's avatar
      scsi: hisi_sas: Remove preemptible() · 18577cdc
      Ahmed S. Darwish authored 
      hisi_sas_task_exec() uses preemptible() to see if it's safe to block.  This
does not work for CONFIG_PREEMPT_COUNT=n kernels in which preemptible()
always returns 0.

The problem is masked when enabling some of the common Kconfig.debug
options (like CONFIG_DEBUG_ATOMIC_SLEEP), as they implicitly enable the
preemption counter.

In general, driver leaf functions should not make logic decisions based on
the context they're called from. The caller should be the entity
responsible for explicitly indicating context.

Since hisi_sas_task_exec() already has a gfp_t flags parameter, use it as
the explicit context marker.

Link: https://lore.kernel.org/r/20201126132952.2287996-3-bigeasy@linutronix.de
Fixes: 214e702d ("scsi: hisi_sas: Adjust task reject period during host reset")
Fixes: 550c0d89 ("scsi: hisi_sas: Replace in_softirq() check in hisi_sas_task_exec()")
Cc: Xiaofei Tan <tanxiaofei@huawei.com>
Cc: Xiang Chen <chenxiang66@hisilicon.com>
Cc: John Garry <john.garry@huawei.com>
Acked-by: default avatarJohn Garry <john.garry@huawei.com>
Signed-off-by: default avatarAhmed S. Darwish <a.darwish@linutronix.de>
Signed-off-by: default avatarSebastian Andrzej Siewior <bigeasy@linutronix.de>
Signed-off-by: default avatarMartin K. Petersen <martin.petersen@oracle.com>
      18577cdc
GitLab Nexedi Edition | About GitLab | About Nexedi | 沪ICP备2021021310号-2 | 沪ICP备2021021310号-7