nsio_rw_bytes can clear media errors, but this cannot be done while we
are in an atomic context due to locking within ACPI. From the BTT,
->rw_bytes may be called either from atomic or process context depending
on whether the calls happen during initialization or during IO.
During init, we want to ensure error clearing happens, and the flag
marking process context allows nsio_rw_bytes to do that. When called
during IO, we're in atomic context, and error clearing can be skipped.
Cc: Dan Williams <dan.j.williams@intel.com>
Signed-off-by: Vishal Verma <vishal.l.verma@intel.com>
Signed-off-by: Dan Williams <dan.j.williams@intel.com>
This continues the 4.11 status quo of disabling of error clearing from
the BTT I/O path. Toshi found that even though we have eliminated all
the libnvdimm sources of sleeping-while-atomic triggers, we still have
sleeping operations that will occur in the path to send the ACPI DSM to
the DIMM to clear the error:
BUG: sleeping function called from invalid context at mm/slab.h:432
in_atomic(): 1, irqs_disabled(): 0, pid: 13353, name: dd
Call Trace:
dump_stack+0x86/0xc3
___might_sleep+0x17d/0x250
__might_sleep+0x4a/0x80
__kmalloc+0x1c0/0x2e0
acpi_os_allocate_zeroed+0x2d/0x2f
acpi_evaluate_object+0x59/0x3b1
acpi_evaluate_dsm+0xbd/0x10c
acpi_nfit_ctl+0x1ef/0x7c0 [nfit]
? nsio_rw_bytes+0x152/0x280
nvdimm_clear_poison+0x77/0x140
nsio_rw_bytes+0x18f/0x280
btt_write_pg+0x1d4/0x3d0 [nd_btt]
btt_make_request+0x119/0x2d0 [nd_btt]
A solution for tracking and handling media errors natively in the BTT is
needed.
Cc: Jeff Moyer <jmoyer@redhat.com>
Cc: Dave Jiang <dave.jiang@intel.com>
Cc: Vishal Verma <vishal.l.verma@intel.com>
Reported-by: Toshi Kani <toshi.kani@hpe.com>
Signed-off-by: Dan Williams <dan.j.williams@intel.com>
A debug patch to turn the standard device_lock() into something that
lockdep can analyze yielded the following:
======================================================
[ INFO: possible circular locking dependency detected ]
4.11.0-rc4+ #106 Tainted: G O
-------------------------------------------------------
lt-libndctl/1898 is trying to acquire lock:
(&dev->nvdimm_mutex/3){+.+.+.}, at: [<ffffffffc023c948>] nd_attach_ndns+0x178/0x1b0 [libnvdimm]
but task is already holding lock:
(&nvdimm_bus->reconfig_mutex){+.+.+.}, at: [<ffffffffc022e0b1>] nvdimm_bus_lock+0x21/0x30 [libnvdimm]
which lock already depends on the new lock.
the existing dependency chain (in reverse order) is:
-> #1 (&nvdimm_bus->reconfig_mutex){+.+.+.}:
lock_acquire+0xf6/0x1f0
__mutex_lock+0x88/0x980
mutex_lock_nested+0x1b/0x20
nvdimm_bus_lock+0x21/0x30 [libnvdimm]
nvdimm_namespace_capacity+0x1b/0x40 [libnvdimm]
nvdimm_namespace_common_probe+0x230/0x510 [libnvdimm]
nd_pmem_probe+0x14/0x180 [nd_pmem]
nvdimm_bus_probe+0xa9/0x260 [libnvdimm]
-> #0 (&dev->nvdimm_mutex/3){+.+.+.}:
__lock_acquire+0x1107/0x1280
lock_acquire+0xf6/0x1f0
__mutex_lock+0x88/0x980
mutex_lock_nested+0x1b/0x20
nd_attach_ndns+0x178/0x1b0 [libnvdimm]
nd_namespace_store+0x308/0x3c0 [libnvdimm]
namespace_store+0x87/0x220 [libnvdimm]
In this case '&dev->nvdimm_mutex/3' mirrors '&dev->mutex'.
Fix this by replacing the use of device_lock() with nvdimm_bus_lock() to protect
nd_{attach,detach}_ndns() operations.
Cc: <stable@vger.kernel.org>
Fixes: 8c2f7e8658 ("libnvdimm: infrastructure for btt devices")
Reported-by: Yi Zhang <yizhan@redhat.com>
Signed-off-by: Dan Williams <dan.j.williams@intel.com>
nvdimm_clear_poison() expects a physical address, not an offset.
Fix nsio_rw_bytes() to call nvdimm_clear_poison() with a physical
address.
Signed-off-by: Toshi Kani <toshi.kani@hpe.com>
Cc: Dave Jiang <dave.jiang@intel.com>
Cc: Vishal Verma <vishal.l.verma@intel.com>
Reviewed-by: Vishal Verma <vishal.l.verma@intel.com>
Signed-off-by: Dan Williams <dan.j.williams@intel.com>
memcpy_from_pmem() maps directly to memcpy_mcsafe(). The wrapper
serves no real benefit aside from affording a more generic function name
than the x86-specific 'mcsafe'. However this would not be the first time
that x86 terminology leaked into the global namespace. For lack of
better name, just use memcpy_mcsafe() directly.
This conversion also catches a place where we should have been using
plain memcpy, acpi_nfit_blk_single_io().
Cc: <x86@kernel.org>
Cc: Jan Kara <jack@suse.cz>
Cc: Jeff Moyer <jmoyer@redhat.com>
Cc: Ingo Molnar <mingo@redhat.com>
Cc: Christoph Hellwig <hch@lst.de>
Cc: "H. Peter Anvin" <hpa@zytor.com>
Cc: Thomas Gleixner <tglx@linutronix.de>
Cc: Matthew Wilcox <mawilcox@microsoft.com>
Cc: Ross Zwisler <ross.zwisler@linux.intel.com>
Acked-by: Tony Luck <tony.luck@intel.com>
Signed-off-by: Dan Williams <dan.j.williams@intel.com>
This reverts commit 4aa5615e08 "libnvdimm: band aid btt vs clear
poison locking".
Now that poison list locking has been converted to a spinlock and poison
list entry allocation during i/o has been converted to GFP_NOWAIT,
revert the band-aid that disabled error clearing from btt i/o.
Cc: Vishal Verma <vishal.l.verma@intel.com>
Cc: Dave Jiang <dave.jiang@intel.com>
Signed-off-by: Dan Williams <dan.j.williams@intel.com>
The following warning results from holding a lane spinlock,
preempt_disable(), or the btt map spinlock and then trying to take the
reconfig_mutex to walk the poison list and potentially add new entries.
BUG: sleeping function called from invalid context at kernel/locking/mutex.c:747
in_atomic(): 1, irqs_disabled(): 0, pid: 17159, name: dd
[..]
Call Trace:
dump_stack+0x85/0xc8
___might_sleep+0x184/0x250
__might_sleep+0x4a/0x90
__mutex_lock+0x58/0x9b0
? nvdimm_bus_lock+0x21/0x30 [libnvdimm]
? __nvdimm_bus_badblocks_clear+0x2f/0x60 [libnvdimm]
? acpi_nfit_forget_poison+0x79/0x80 [nfit]
? _raw_spin_unlock+0x27/0x40
mutex_lock_nested+0x1b/0x20
nvdimm_bus_lock+0x21/0x30 [libnvdimm]
nvdimm_forget_poison+0x25/0x50 [libnvdimm]
nvdimm_clear_poison+0x106/0x140 [libnvdimm]
nsio_rw_bytes+0x164/0x270 [libnvdimm]
btt_write_pg+0x1de/0x3e0 [nd_btt]
? blk_queue_enter+0x30/0x290
btt_make_request+0x11a/0x310 [nd_btt]
? blk_queue_enter+0xb7/0x290
? blk_queue_enter+0x30/0x290
generic_make_request+0x118/0x3b0
As a minimal fix, disable error clearing when the BTT is enabled for the
namespace. For the final fix a larger rework of the poison list locking
is needed.
Note that this is not a problem in the blk case since that path never
calls nvdimm_clear_poison().
Cc: <stable@vger.kernel.org>
Fixes: 82bf1037f2 ("libnvdimm: check and clear poison before writing to pmem")
Cc: Dave Jiang <dave.jiang@intel.com>
[jeff: dynamically disable error clearing in the btt case]
Suggested-by: Jeff Moyer <jmoyer@redhat.com>
Reviewed-by: Jeff Moyer <jmoyer@redhat.com>
Reported-by: Vishal Verma <vishal.l.verma@intel.com>
Signed-off-by: Dan Williams <dan.j.williams@intel.com>
Colin, via static analysis, reports that the length could be negative
from nvdimm_clear_poison() in the error case. There was a similar
problem with commit 0a3f27b9a6 "libnvdimm, namespace: avoid multiple
sector calculations" that I noticed when merging the for-4.10/libnvdimm
topic branch into libnvdimm-for-next, but I missed this one. Fix both of
them to the following procedure:
* if we clear a block's worth of media, clear that many blocks in
badblocks
* if we clear less than the requested size of the transfer return an
error
* always invalidate cache after any non-error / non-zero
nvdimm_clear_poison result
Fixes: 82bf1037f2 ("libnvdimm: check and clear poison before writing to pmem")
Fixes: 0a3f27b9a6 ("libnvdimm, namespace: avoid multiple sector calculations")
Cc: Fabian Frederick <fabf@skynet.be>
Cc: Dave Jiang <dave.jiang@intel.com>
Reported-by: Colin Ian King <colin.king@canonical.com>
Signed-off-by: Dan Williams <dan.j.williams@intel.com>
For warnings that should only ever trigger during development and
testing replace WARN statements with lockdep_assert_held. The lockdep
pattern is prevalent, and these paths are are well covered by libnvdimm
unit tests.
Reported-by: Johannes Thumshirn <jthumshirn@suse.de>
Reviewed-by: Johannes Thumshirn <jthumshirn@suse.de>
Signed-off-by: Dan Williams <dan.j.williams@intel.com>
else after return is not needed.
Signed-off-by: Fabian Frederick <fabf@skynet.be>
[djbw: removed some now unnecessary newlines]
Signed-off-by: Dan Williams <dan.j.williams@intel.com>
Here is an example /proc/iomem listing for a system with 2 namespaces,
one in "sector" mode and one in "memory" mode:
1fc000000-2fbffffff : Persistent Memory (legacy)
1fc000000-2fbffffff : namespace1.0
340000000-34fffffff : Persistent Memory
340000000-34fffffff : btt0.1
Here is the corresponding ndctl listing:
# ndctl list
[
{
"dev":"namespace1.0",
"mode":"memory",
"size":4294967296,
"blockdev":"pmem1"
},
{
"dev":"namespace0.0",
"mode":"sector",
"size":267091968,
"uuid":"f7594f86-badb-4592-875f-ded577da2eaf",
"sector_size":4096,
"blockdev":"pmem0s"
}
]
Notice that the ndctl listing is purely in terms of namespace devices,
while the iomem listing leaks the internal "btt0.1" implementation
detail. Given that ndctl requires the namespace device name to change
the mode, for example:
# ndctl create-namespace --reconfig=namespace0.0 --mode=raw --force
...use the namespace name in the iomem listing to keep the claiming
device name consistent across different mode settings.
Cc: Vishal Verma <vishal.l.verma@intel.com>
Signed-off-by: Dan Williams <dan.j.williams@intel.com>
We need to clear any poison when we are writing to pmem. The granularity
will be sector size. If it's less then we can't do anything about it
barring corruption.
Signed-off-by: Dave Jiang <dave.jiang@intel.com>
Reviewed-by: Vishal Verma <vishal.l.verma@intel.com>
[djbw: fixup 0-length write request to succeed]
Signed-off-by: Dan Williams <dan.j.williams@intel.com>
nsio_rw_bytes() is used to write info block metadata to the namespace,
so it should trigger a flush after every write. Replace wmb_pmem() with
nvdimm_flush() in this path.
Cc: Ross Zwisler <ross.zwisler@linux.intel.com>
Signed-off-by: Dan Williams <dan.j.williams@intel.com>
Prompted by commit 287980e49f "remove lots of IS_ERR_VALUE abuses", I
ran make coccicheck against drivers/nvdimm/ and found that:
if (IS_ERR(x))
return PTR_ERR(x);
return 0;
...can be replaced with PTR_ERR_OR_ZERO().
Reported-by: Linus Torvalds <torvalds@linux-foundation.org>
Reviewed-by: Johannes Thumshirn <jthumshirn@suse.de>
Signed-off-by: Dan Williams <dan.j.williams@intel.com>
The ndctl unit tests discovered that the dax enabling omitted updates to
nd_detach_and_reset(). This routine clears device the configuration
when the namespace is detached. Without this clearing userspace may
assume that the device is in the process of being configured by another
agent in the system.
Signed-off-by: Dan Williams <dan.j.williams@intel.com>
Device DAX is the device-centric analogue of Filesystem DAX
(CONFIG_FS_DAX). It allows persistent memory ranges to be allocated and
mapped without need of an intervening file system. This initial
infrastructure arranges for a libnvdimm pfn-device to be represented as
a different device-type so that it can be attached to a driver other
than the pmem driver.
Signed-off-by: Dan Williams <dan.j.williams@intel.com>
In preparation for providing an alternative (to block device) access
mechanism to persistent memory, convert pmem_rw_bytes() to
nsio_rw_bytes(). This allows ->rw_bytes() functionality without
requiring a 'struct pmem_device' to be instantiated.
In other words, when ->rw_bytes() is in use i/o is driven through
'struct nd_namespace_io', otherwise it is driven through 'struct
pmem_device' and the block layer. This consolidates the disjoint calls
to devm_exit_badblocks() and devm_memunmap() into a common
devm_nsio_disable() and cleans up the init path to use a unified
pmem_attach_disk() implementation.
Reviewed-by: Johannes Thumshirn <jthumshirn@suse.de>
Signed-off-by: Dan Williams <dan.j.williams@intel.com>
Implement the base infrastructure for libnvdimm PFN devices. Similar to
BTT devices they take a namespace as a backing device and layer
functionality on top. In this case the functionality is reserving space
for an array of 'struct page' entries to be handed out through
pfn_to_page(). For now this is just the basic libnvdimm-device-model for
configuring the base PFN device.
As the namespace claiming mechanism for PFN devices is mostly identical
to BTT devices drivers/nvdimm/claim.c is created to house the common
bits.
Cc: Ross Zwisler <ross.zwisler@linux.intel.com>
Signed-off-by: Dan Williams <dan.j.williams@intel.com>
