- Jul 12, 2022
-
-
Greg Kroah-Hartman authored
Merge tag 'misc-habanalabs-next-2022-07-12' of https://git.kernel.org/pub/scm/linux/kernel/git/ogabbay/linux into char-misc-next Oded writes: This tag contains habanalabs driver changes for v5.20: - Add Gaudi2 ASIC support. All the features required for Gaudi2 are included in this tag (except the networking aspect). - Add more events to the eventfd support in the driver. With the new code, we expose three events that the user can register to get notification about them. - re-factor soft reset code and replace its name to compute reset to better reflect the actual reset done in new ASICs - Change the way Gaudi2 triggers an MSI-X interrupt due to h/w bug. - Improve the code of the debugfs node that scrubs the device's memory. - Add mechanism for better compatibility with older f/w versions - Cleanup kernel log prints by moving some prints to debug and removing others. - Many small bug fixes and minor changes. * tag 'misc-habanalabs-next-2022-07-12' of https://git.kernel.org/pub/scm/linux/kernel/git/ogabbay/linux: (88 commits) habanalabs: move h/w dirty message to debug habanalabs: rename soft reset to compute reset habanalabs: add status of reset after device release habanalabs: fix update of is_in_soft_reset habanalabs: expose only valid debugfs nodes habanalabs/gaudi2: map virtual MSI-X doorbell memory for user habanalabs/gaudi2: modify decoder to use virtual MSI-X doorbell habanalabs/gaudi2: modify CS completion CQ to use virtual MSI-X doorbell habanalabs/gaudi2: replace defines for reserved sob/mob with enums habanalabs/gaudi2: configure virtual MSI-X doorbell interface habanalabs: add a value field to hl_fw_send_pci_access_msg() habanalabs: fixes to the poll-timeout macros habanalabs/gaudi2: use DIV_ROUND_UP_SECTOR_T instead of roundup habanalabs: initialize variable explicitly habanalabs: Use the bitmap API to allocate bitmaps habanalabs/gaudi2: remove unused defines habanalabs: make sure variable is set before used habanalabs: don't declare tmp twice in same function habanalabs: do not set max power on a secured device habanalabs/gaudi2: SM mask can only be 8-bit ...
-
keliu authored
Use ida_alloc()/ida_free() instead of deprecated ida_simple_get()/ida_simple_remove() . Signed-off-by:
keliu <liuke94@huawei.com> Signed-off-by:
Johannes Thumshirn <jth@kernel.org> Link: https://lore.kernel.org/r/480676bee970da16bf1fa8565277240014395ba3.1657607743.git.johannes.thumshirn@wdc.com Signed-off-by:
Greg Kroah-Hartman <gregkh@linuxfoundation.org>
-
Oded Gabbay authored
H/W being dirty during initialization is completely expected in case f/w tools are used before loading the driver. As it is not an error, and as it doesn't give any meaningful information to the user, no point of printing it. Signed-off-by:Oded Gabbay <ogabbay@kernel.org>
-
Oded Gabbay authored
Doing compute reset can be the traditional inference soft reset that is supported only in Goya. Or it can be the new reset upon device release, which is supported in Gaudi2 and above. Therefore, wherever suitable, use the terminology of compute reset instead of soft reset. Signed-off-by: -
Oded Gabbay authored
The user might want to know the device is in reset after device release, which is not an erroneous event as a regular reset. Signed-off-by: -
Oded Gabbay authored
reset_info.is_in_soft_reset should be updated both before in_reset and inside the spin lock of the reset info structure. The reasons are: - When we are inside soft reset, it implies we are in reset. Therefore, if someone checks if we are in soft reset, he can deduce we are in reset, while the opposite is not correct and might be misleading. - Both these flags are changed together so they must be changed inside the reset info spinlock. Signed-off-by: -
Ofir Bitton authored
In case security is enabled on the device, some debugfs nodes will fail. Hence, we do not expose them. Signed-off-by:
Ofir Bitton <obitton@habana.ai> Reviewed-by:
-
Tomer Tayar authored
Upon the initialization of a user context, map the host memory page of the virtual MSI-X doorbell in the device MMU. A reserved VA is used for this purpose, so user can use it directly without any allocation/map operation. Signed-off-by:
Tomer Tayar <ttayar@habana.ai> Reviewed-by:
-
Tomer Tayar authored
Modify the decoder wrapper blocks to generate interrupts using the virtual MSI-X doorbell. As a decoder wrapper block cannot write directly to HBW upon completion, it writes instead to SOB which is monitored by a master monitor. When resolved, this monitor will be the one to actually write to the virtual MSI-X doorbell. Signed-off-by:
-
Tomer Tayar authored
Modify the CQ which is used for CS completion, to use the virtual MSI-X doorbell. Signed-off-by:
-
Tomer Tayar authored
Following patches are going to add more reserved sync objects and monitors. To make the counting of these reserved resources simpler, replace the existing RESERVED_* defines with enumerations. Signed-off-by:
-
Tomer Tayar authored
Due to a watchdog timer in the LBW path, writes to the MSI-X doorbell can return sporadic error responses. To work-around this issue, a virtual MSI-X doorbell on the HBW path is configured, using the MSI-X AXI slave interface in the PCIe controller. Upon an access to a configured HBW host address, the controller will generate MSI-X interrupt instead of treating the access as regular host memory access. This patch allocates the dedicate host memory page, and communicate the address to F/W, so it will configure the relevant address match registers in the controller, and will use this address to generate MSI-X interrupts for F/W events. Following patches will handle other initiators in the device, to move them to use the virtual MSI-X doorbell. Signed-off-by:
-
Tomer Tayar authored
For gaudi2 we need to send a value to F/W as part of the PCI_ACCESS packet. As a preparation, modify hl_fw_send_pci_access_msg() to have a 'value' field. Signed-off-by:
-
Ohad Sharabi authored
- use conventional internal macro variables (double underscore prefix) - adjust address casting - on register poll using ELBI use ELBI read rather than BAR read on error condition - remove unused macro Signed-off-by:
Ohad Sharabi <osharabi@habana.ai> Reviewed-by:
-
Ohad Sharabi authored
roundup will create an error in 32-bit architectures as we use 64-bit variables. Reported-by:
kernel test robot <lkp@intel.com> Signed-off-by:
-
Oded Gabbay authored
Fix warning of "warning: ‘old_base’ may be used uninitialized in this function" Signed-off-by: -
Christophe JAILLET authored
Use bitmap_zalloc()/bitmap_free() instead of hand-writing them. It is less verbose and it improves the semantic. Signed-off-by:
Christophe JAILLET <christophe.jaillet@wanadoo.fr> Reviewed-by:
-
Oded Gabbay authored
There were some defines that are unused in the current upstreamed code. Signed-off-by: -
Oded Gabbay authored
timestamp could be unset in both _hl_interrupt_wait_ioctl() and _hl_interrupt_wait_ioctl_user_addr() so it is better to explicitly initialize it to 0 when declaring it. Signed-off-by: -
Oded Gabbay authored
tmp is declared in the scope of the function cs_do_release() and inside a block inside that function. Signed-off-by: -
Ofir Bitton authored
Max power API is not supported in secured devices. Hence, we should skip setting it during boot. Signed-off-by:
-
Oded Gabbay authored
Otherwise, due to how we calculate it, we might fail in FIELD_PREP checks. Reported-by:
-
Oded Gabbay authored
glbl_sts_clr_val was set but never used Reported-by:
-
Oded Gabbay authored
If we send a packet to the f/w, and that packet is unsupported, we want to be able to identify this situation and possibly ignore this. Therefore, if the f/w returned an error, we need to propagate it to the callers in the result value, if those callers were interested in it. In addition, no point of printing the error code here because each caller prints its own error with a specific message. Signed-off-by: -
Sagiv Ozeri authored
We need this property for backward compatibility against the f/w. Signed-off-by:
Sagiv Ozeri <sozeri@habana.ai> Reviewed-by:
-
Ofir Bitton authored
User application should be able to get notification for any decoder completion. Hence, we introduce a new interface in which a user can wait for all current decoder pending interrupts. Signed-off-by:
-
Ofir Bitton authored
Current naming convention can be misleading. Hence renaming some variables and defines in order to be more explicit. Signed-off-by:
-
Ohad Sharabi authored
Currently we are not waiting for preboot ready after hard reset. This leads to a race in which COMMs protocol begins but will get no response from the f/w. Signed-off-by:
-
Ofir Bitton authored
Correctable ECC events are not fatal, but as they accumulate, the f/w can decide that a hard-rest is required. This indication is propagated to the host using the existing ECC event interface. Signed-off-by:
-
Oded Gabbay authored
Enable the Gaudi2 ASIC code in the pci probe callback of the driver so the driver will handle Gaudi2 ASICs. Add the PCI ID to the PCI table and add the ASIC enum value to all relevant places. Fixup the device parameters initialization for Gaudi2. Signed-off-by: -
Moti Haimovski authored
Gaudi2 has new MMU units. A PMMU for device->host accesses, and HMMU for HBM accesses. The page tables of both MMUs are located in the host's memory (referred to in the code as host-resident pgt). Signed-off-by:
Moti Haimovski <mhaimovski@habana.ai> Reviewed-by:
-
Oded Gabbay authored
In Gaudi2 we moved to a different wait for command submission completion model. Instead of receiving interrupt only on external queues, we use the device's sync manager to notify us when the entire command submission finishes. This enables us to remove the categorization of queues to external and internal, and treat each queue equally, without the need to parse and patch any command buffer. This change also requires refactoring to the IRQ handling of CS completions. Signed-off-by: -
Benjamin Dotan authored
Add the Gaudi2 code to initialize the ASIC's profiler. The profile receives its initialization values from the user, same as in Gaudi2, but the code to initialize is in the driver because the configuration space of the device is not directly exposed to the user. Signed-off-by:
Benjamin Dotan <bdotan@habana.ai> Reviewed-by:
-
Ofir Bitton authored
Use the generic security module to block all registers in the ASIC and then open only those that are needed to be accessed by the user. Signed-off-by:
-
Ofir Bitton authored
As the ASICs become more complex and have many more registers, we need a better way to configure the security properties. As a reminder, we have two dedicated mechanisms for security: Range Registers and Protection bits. Those mechanisms protect sensitive memory and configuration areas inside the device. The generic module handles the low-level part of the configuration, because the configuration mechanism is identical in all ASICs. The difference is the address ranges and register names. Any ASIC that use this block should first block all the register blocks in the ASIC. Then, it should open only the registers that need to be accessed by the user (This is opposed to Goya and Gaudi, where we blocked only what should not be accesses by the user). The module contains several functions, to unblock single register, multiple registers, entire blocks, ranges, ranges with mask. Signed-off-by:
-
Oded Gabbay authored
There are a couple of device variables that are used for testing purposes and they are set to fixed values. Remove the variables that are not relevant anymore and document the remaining variables. Signed-off-by: -
Oded Gabbay authored
New asic properties were added for Gaudi2. We want to initialize and use them, when relevant, also for Goya and Gaudi. Signed-off-by: -
Oded Gabbay authored
There are a number of new ASIC-specific functions that were added for Gaudi2. To make the common code work, we need to define empty implementations of those functions for Goya and Gaudi. Some functions will return error if called with Goya/Gaudi. Signed-off-by: -
Oded Gabbay authored
Add the ASIC-specific code for Gaudi2. Supply (almost) all of the function callbacks that the driver's common code need to initialize, finalize and submit workloads to the Gaudi2 ASIC. It also contains the code to initialize the F/W of the Gaudi2 ASIC and to receive events from the F/W. It contains new debugfs entry to dump razwi events. razwi is a case where the device's engines create a transaction that reaches an invalid destination. Signed-off-by: -
Oded Gabbay authored
Add the new defines for GAUDI2 uapi interface. It includes the following: 1. Enums of engines and PLLs. 2. New information in the info IOCTL that is retrieved by the driver. 3. Update comments regarding the CB/CS/wait for CS ioctls. 4. New fields in the debug IOCTL for configuring the profiler for Gaudi2. There is no new IOCTL. Some of the changes are also relevant for Greco (which will be upstreamed later this year). When ever it says "Greco and onwards", it means it is also for Gaudi2. Signed-off-by:
-
