linux-stable.git/drivers/crypto/ccp, branch linux-6.3.y

crypto: ccp - Clear PSP interrupt status register before calling handler

2023-05-17T12:01:35+00:00

[ Upstream commit 45121ad4a1750ca47ce3f32bd434bdb0cdbf0043 ]

The PSP IRQ is edge-triggered (MSI or MSI-X) in all cases supported by
the psp module so clear the interrupt status register early in the
handler to prevent missed interrupts. sev_irq_handler() calls wake_up()
on a wait queue, which can result in a new command being submitted from
a different CPU. This then races with the clearing of isr and can result
in missed interrupts. A missed interrupt results in a command waiting
until it times out, which results in the psp being declared dead.

This is unlikely on bare metal, but has been observed when running
virtualized. In the cases where this is observed, sev->cmdresp_reg has
PSP_CMDRESP_RESP set which indicates that the command was processed
correctly but no interrupt was asserted.

The full sequence of events looks like this:

CPU 1: submits SEV cmd #1
CPU 1: calls wait_event_timeout()
CPU 0: enters psp_irq_handler()
CPU 0: calls sev_handler()->wake_up()
CPU 1: wakes up; finishes processing cmd #1
CPU 1: submits SEV cmd #2
CPU 1: calls wait_event_timeout()
PSP:   finishes processing cmd #2; interrupt status is still set; no interrupt
CPU 0: clears intsts
CPU 0: exits psp_irq_handler()
CPU 1: wait_event_timeout() times out; psp_dead=true

Fixes: 200664d5237f ("crypto: ccp: Add Secure Encrypted Virtualization (SEV) command support")
Cc: stable@vger.kernel.org
Signed-off-by: Jeremi Piotrowski 
Acked-by: Tom Lendacky 
Signed-off-by: Herbert Xu 
Signed-off-by: Sasha Levin

crypto: ccp - Don't initialize CCP for PSP 0x1649

2023-05-11T14:16:50+00:00

commit c79a3169b9f3633c215b55857eba5921e5b49217 upstream.

A number of platforms are emitting the error:
```ccp: unable to access the device: you might be running a broken BIOS.```

This is expected behavior as CCP is no longer accessible from the PSP's
PCIe BAR so stop trying to probe CCP for 0x1649.

Cc: stable@vger.kernel.org
Signed-off-by: Mario Limonciello 
Acked-by: Tom Lendacky 
Signed-off-by: Herbert Xu 
Signed-off-by: Greg Kroah-Hartman

crypto: ccp - Use request_complete helpers

2023-02-13T10:34:49+00:00

Use the request_complete helpers instead of calling the completion
function directly.

Signed-off-by: Herbert Xu 
Acked-by: Tom Lendacky 
Signed-off-by: Herbert Xu

crypto: ccp - Flush the SEV-ES TMR memory before giving it to firmware

2023-02-03T04:54:54+00:00

Perform a cache flush on the SEV-ES TMR memory after allocation to prevent
any possibility of the firmware encountering an error should dirty cache
lines be present. Use clflush_cache_range() to flush the SEV-ES TMR memory.

Fixes: 97f9ac3db661 ("crypto: ccp - Add support for SEV-ES to the PSP driver")
Signed-off-by: Tom Lendacky 
Signed-off-by: Herbert Xu

crypto: ccp - Add a firmware definition for EPYC gen 4 processors

2023-01-27T10:57:24+00:00

Add a new MODULE_FIRMWARE() entry for 4th generation EPYC processors.

Signed-off-by: Tom Lendacky 
Signed-off-by: Herbert Xu

crypto: ccp - Provide MMIO register naming for documenation

2023-01-27T10:57:24+00:00

Add comments next to the version data MMIO register values to identify
the register name being used.

Signed-off-by: Tom Lendacky 
Signed-off-by: Herbert Xu

crypto: ccp - Failure on re-initialization due to duplicate sysfs filename

2023-01-20T10:29:31+00:00

The following warning appears during the CCP module re-initialization:

[  140.965403] sysfs: cannot create duplicate filename
'/devices/pci0000:00/0000:00:07.1/0000:03:00.2/dma/dma0chan0'
[  140.975736] CPU: 0 PID: 388 Comm: kworker/0:2 Kdump: loaded Not
tainted 6.2.0-0.rc2.18.eln124.x86_64 #1
[  140.985185] Hardware name: HPE ProLiant DL325 Gen10/ProLiant DL325
Gen10, BIOS A41 07/17/2020
[  140.993761] Workqueue: events work_for_cpu_fn
[  140.998151] Call Trace:
[  141.000613]  
[  141.002726]  dump_stack_lvl+0x33/0x46
[  141.006415]  sysfs_warn_dup.cold+0x17/0x23
[  141.010542]  sysfs_create_dir_ns+0xba/0xd0
[  141.014670]  kobject_add_internal+0xba/0x260
[  141.018970]  kobject_add+0x81/0xb0
[  141.022395]  device_add+0xdc/0x7e0
[  141.025822]  ? complete_all+0x20/0x90
[  141.029510]  __dma_async_device_channel_register+0xc9/0x130
[  141.035119]  dma_async_device_register+0x19e/0x3b0
[  141.039943]  ccp_dmaengine_register+0x334/0x3f0 [ccp]
[  141.045042]  ccp5_init+0x662/0x6a0 [ccp]
[  141.049000]  ? devm_kmalloc+0x40/0xd0
[  141.052688]  ccp_dev_init+0xbb/0xf0 [ccp]
[  141.056732]  ? __pci_set_master+0x56/0xd0
[  141.060768]  sp_init+0x70/0x90 [ccp]
[  141.064377]  sp_pci_probe+0x186/0x1b0 [ccp]
[  141.068596]  local_pci_probe+0x41/0x80
[  141.072374]  work_for_cpu_fn+0x16/0x20
[  141.076145]  process_one_work+0x1c8/0x380
[  141.080181]  worker_thread+0x1ab/0x380
[  141.083953]  ? __pfx_worker_thread+0x10/0x10
[  141.088250]  kthread+0xda/0x100
[  141.091413]  ? __pfx_kthread+0x10/0x10
[  141.095185]  ret_from_fork+0x2c/0x50
[  141.098788]  
[  141.100996] kobject_add_internal failed for dma0chan0 with -EEXIST,
don't try to register things with the same name in the same directory.
[  141.113703] ccp 0000:03:00.2: ccp initialization failed

The /dma/dma0chan0 sysfs file is not removed since dma_chan object
has been released in ccp_dma_release() before releasing dma device.
A correct procedure would be: release dma channels first => unregister
dma device => release ccp dma object.

Link: https://bugzilla.kernel.org/show_bug.cgi?id=216888
Fixes: 68dbe80f5b51 ("crypto: ccp - Release dma channels before dmaengine unrgister")
Tested-by: Vladis Dronov 
Signed-off-by: Koba Ko 
Reviewed-by: Vladis Dronov 
Acked-by: Tom Lendacky 
Signed-off-by: Herbert Xu

crypto: ccp - Avoid page allocation failure warning for SEV_GET_ID2

2023-01-06T09:15:47+00:00

For SEV_GET_ID2, the user provided length does not have a specified
limitation because the length of the ID may change in the future.  The
kernel memory allocation, however, is implicitly limited to 4MB on x86 by
the page allocator, otherwise the kzalloc() will fail.

When this happens, it is best not to spam the kernel log with the warning.
Simply fail the allocation and return ENOMEM to the user.

Fixes: d6112ea0cb34 ("crypto: ccp - introduce SEV_GET_ID2 command")
Reported-by: Andy Nguyen 
Reported-by: Peter Gonda 
Suggested-by: Herbert Xu 
Signed-off-by: David Rientjes 
Acked-by: Tom Lendacky 
Signed-off-by: Herbert Xu

crypto: ccp - Set DMA alignment explicitly

2022-12-09T10:45:00+00:00

This driver has been implicitly relying on kmalloc alignment
to be sufficient for DMA.  This may no longer be the case with
upcoming arm64 changes.

This patch changes it to explicitly request DMA alignment from
the Crypto API.

Signed-off-by: Herbert Xu

crypto: ccp - Add support for TEE for PCI ID 0x14CA

2022-10-21T11:15:35+00:00

SoCs containing 0x14CA are present both in datacenter parts that
support SEV as well as client parts that support TEE.

Cc: stable@vger.kernel.org # 5.15+
Tested-by: Rijo-john Thomas 
Signed-off-by: Mario Limonciello 
Acked-by: Tom Lendacky 
Signed-off-by: Herbert Xu