linux-stable.git/drivers/edac, branch linux-4.9.y

EDAC: Fix calculation of returned address and next offset in edac_align_ptr()

2022-02-23T10:56:41+00:00

commit f8efca92ae509c25e0a4bd5d0a86decea4f0c41e upstream.

Do alignment logic properly and use the "ptr" local variable for
calculating the remainder of the alignment.

This became an issue because struct edac_mc_layer has a size that is not
zero modulo eight, and the next offset that was prepared for the private
data was unaligned, causing an alignment exception.

The patch in Fixes: which broke this actually wanted to "what we
actually care about is the alignment of the actual pointer that's about
to be returned." But it didn't check that alignment.

Use the correct variable "ptr" for that.

  [ bp: Massage commit message. ]

Fixes: 8447c4d15e35 ("edac: Do alignment logic properly in edac_align_ptr()")
Signed-off-by: Eliav Farber 
Signed-off-by: Borislav Petkov 
Cc: 
Link: https://lore.kernel.org/r/20220113100622.12783-2-farbere@amazon.com
Signed-off-by: Greg Kroah-Hartman

EDAC/xgene: Fix deferred probing

2022-02-08T17:15:30+00:00

commit dfd0dfb9a7cc04acf93435b440dd34c2ca7b4424 upstream.

The driver overrides error codes returned by platform_get_irq_optional()
to -EINVAL for some strange reason, so if it returns -EPROBE_DEFER, the
driver will fail the probe permanently instead of the deferred probing.
Switch to propagating the proper error codes to platform driver code
upwards.

  [ bp: Massage commit message. ]

Fixes: 0d4429301c4a ("EDAC: Add APM X-Gene SoC EDAC driver")
Signed-off-by: Sergey Shtylyov 
Signed-off-by: Borislav Petkov 
Cc: 
Link: https://lore.kernel.org/r/20220124185503.6720-3-s.shtylyov@omp.ru
Signed-off-by: Greg Kroah-Hartman

EDAC/altera: Fix deferred probing

2022-02-08T17:15:30+00:00

commit 279eb8575fdaa92c314a54c0d583c65e26229107 upstream.

The driver overrides the error codes returned by platform_get_irq() to
-ENODEV for some strange reason, so if it returns -EPROBE_DEFER, the
driver will fail the probe permanently instead of the deferred probing.
Switch to propagating the proper error codes to platform driver code
upwards.

  [ bp: Massage commit message. ]

Fixes: 71bcada88b0f ("edac: altera: Add Altera SDRAM EDAC support")
Signed-off-by: Sergey Shtylyov 
Signed-off-by: Borislav Petkov 
Acked-by: Dinh Nguyen 
Cc: 
Link: https://lore.kernel.org/r/20220124185503.6720-2-s.shtylyov@omp.ru
Signed-off-by: Greg Kroah-Hartman

EDAC/sb_edac: Fix top-of-high-memory value for Broadwell/Haswell

2021-11-26T10:48:22+00:00

commit 537bddd069c743759addf422d0b8f028ff0f8dbc upstream.

The computation of TOHM is off by one bit. This missed bit results in
too low a value for TOHM, which can cause errors in regular memory to
incorrectly report:

  EDAC MC0: 1 CE Error at MMIOH area, on addr 0x000000207fffa680 on any memory

Fixes: 50d1bb93672f ("sb_edac: add support for Haswell based systems")
Cc: stable@vger.kernel.org
Reported-by: Meeta Saggi 
Signed-off-by: Eric Badger 
Signed-off-by: Tony Luck 
Link: https://lore.kernel.org/r/20211010170127.848113-1-ebadger@purestorage.com
Signed-off-by: Greg Kroah-Hartman

EDAC/synopsys: Fix wrong value type assignment for edac_mode

2021-10-06T08:23:42+00:00

commit 5297cfa6bdf93e3889f78f9b482e2a595a376083 upstream.

dimm->edac_mode contains values of type enum edac_type - not the
corresponding capability flags. Fix that.

Issue caught by Coverity check "enumerated type mixed with another
type."

 [ bp: Rewrite commit message, add tags. ]

Fixes: ae9b56e3996d ("EDAC, synps: Add EDAC support for zynq ddr ecc controller")
Signed-off-by: Sai Krishna Potthuri 
Signed-off-by: Shubhrajyoti Datta 
Signed-off-by: Borislav Petkov 
Cc: 
Link: https://lkml.kernel.org/r/20210818072315.15149-1-shubhrajyoti.datta@xilinx.com
Signed-off-by: Greg Kroah-Hartman

EDAC/i5100: Fix error handling order in i5100_init_one()

2020-10-29T08:05:32+00:00

[ Upstream commit 857a3139bd8be4f702c030c8ca06f3fd69c1741a ]

When pci_get_device_func() fails, the driver doesn't need to execute
pci_dev_put(). mci should still be freed, though, to prevent a memory
leak. When pci_enable_device() fails, the error injection PCI device
"einj" doesn't need to be disabled either.

 [ bp: Massage commit message, rename label to "bail_mc_free". ]

Fixes: 52608ba205461 ("i5100_edac: probe for device 19 function 0")
Signed-off-by: Dinghao Liu 
Signed-off-by: Borislav Petkov 
Link: https://lkml.kernel.org/r/20200826121437.31606-1-dinghao.liu@zju.edu.cn
Signed-off-by: Sasha Levin

EDAC/ie31200: Fallback if host bridge device is already initialized

2020-09-03T09:21:18+00:00

[ Upstream commit 709ed1bcef12398ac1a35c149f3e582db04456c2 ]

The Intel uncore driver may claim some of the pci ids from ie31200 which
means that the ie31200 edac driver will not initialize them as part of
pci_register_driver().

Let's add a fallback for this case to 'pci_get_device()' to get a
reference on the device such that it can still be configured. This is
similar in approach to other edac drivers.

Signed-off-by: Jason Baron 
Cc: Borislav Petkov 
Cc: Mauro Carvalho Chehab 
Cc: linux-edac 
Signed-off-by: Tony Luck 
Link: https://lore.kernel.org/r/1594923911-10885-1-git-send-email-jbaron@akamai.com
Signed-off-by: Sasha Levin

EDAC: Fix reference count leaks

2020-08-21T09:01:56+00:00

[ Upstream commit 17ed808ad243192fb923e4e653c1338d3ba06207 ]

When kobject_init_and_add() returns an error, it should be handled
because kobject_init_and_add() takes a reference even when it fails. If
this function returns an error, kobject_put() must be called to properly
clean up the memory associated with the object.

Therefore, replace calling kfree() and call kobject_put() and add a
missing kobject_put() in the edac_device_register_sysfs_main_kobj()
error path.

 [ bp: Massage and merge into a single patch. ]

Fixes: b2ed215a3338 ("Kobject: change drivers/edac to use kobject_init_and_add")
Signed-off-by: Qiushi Wu 
Signed-off-by: Borislav Petkov 
Link: https://lkml.kernel.org/r/20200528202238.18078-1-wu000273@umn.edu
Link: https://lkml.kernel.org/r/20200528203526.20908-1-wu000273@umn.edu
Signed-off-by: Sasha Levin

EDAC/amd64: Read back the scrub rate PCI register on F15h

2020-07-09T07:35:55+00:00

[ Upstream commit ee470bb25d0dcdf126f586ec0ae6dca66cb340a4 ]

Commit:

  da92110dfdfa ("EDAC, amd64_edac: Extend scrub rate support to F15hM60h")

added support for F15h, model 0x60 CPUs but in doing so, missed to read
back SCRCTRL PCI config register on F15h CPUs which are *not* model
0x60. Add that read so that doing

  $ cat /sys/devices/system/edac/mc/mc0/sdram_scrub_rate

can show the previously set DRAM scrub rate.

Fixes: da92110dfdfa ("EDAC, amd64_edac: Extend scrub rate support to F15hM60h")
Reported-by: Anders Andersson 
Signed-off-by: Borislav Petkov 
Cc:  #v4.4..
Link: https://lkml.kernel.org/r/CAKkunMbNWppx_i6xSdDHLseA2QQmGJqj_crY=NF-GZML5np4Vw@mail.gmail.com
Signed-off-by: Sasha Levin

EDAC/ghes: Fix grain calculation

2020-01-04T12:39:51+00:00

[ Upstream commit 7088e29e0423d3195e09079b4f849ec4837e5a75 ]

The current code to convert a physical address mask to a grain
(defined as granularity in bytes) is:

	e->grain = ~(mem_err->physical_addr_mask & ~PAGE_MASK);

This is broken in several ways:

1) It calculates to wrong grain values. E.g., a physical address mask
of ~0xfff should give a grain of 0x1000. Without considering
PAGE_MASK, there is an off-by-one. Things are worse when also
filtering it with ~PAGE_MASK. This will calculate to a grain with the
upper bits set. In the example it even calculates to ~0.

2) The grain does not depend on and is unrelated to the kernel's
page-size. The page-size only matters when unmapping memory in
memory_failure(). Smaller grains are wrongly rounded up to the
page-size, on architectures with a configurable page-size (e.g. arm64)
this could round up to the even bigger page-size of the hypervisor.

Fix this with:

	e->grain = ~mem_err->physical_addr_mask + 1;

The grain_bits are defined as:

	grain = 1 << grain_bits;

Change also the grain_bits calculation accordingly, it is the same
formula as in edac_mc.c now and the code can be unified.

The value in ->physical_addr_mask coming from firmware is assumed to
be contiguous, but this is not sanity-checked. However, in case the
mask is non-contiguous, a conversion to grain_bits effectively
converts the grain bit mask to a power of 2 by rounding it up.

Suggested-by: James Morse 
Signed-off-by: Robert Richter 
Signed-off-by: Borislav Petkov 
Reviewed-by: Mauro Carvalho Chehab 
Cc: "linux-edac@vger.kernel.org" 
Cc: Tony Luck 
Link: https://lkml.kernel.org/r/20191106093239.25517-11-rrichter@marvell.com
Signed-off-by: Sasha Levin