On Tue, Jun 09, 2020 at 05:15:34AM -0700, Tom Rix wrote:
> When i say 'surprise the user' i mean the end user of the fpga. I do not
> mean the developer of the fpga. An example of the use case..
>
> An AFU to do bitcoin mining is created with a broken but undetected private
> feature in 2017
When i say 'surprise the user' i mean the end user of the fpga. I do not mean
the developer of the fpga. An example of the use case..
An AFU to do bitcoin mining is created with a broken but undetected private
feature in 2017 and starts shipping 10,000+ products/year and runs successfully
On Mon, Jun 08, 2020 at 05:48:22PM -0700, Tom Rix wrote:
> I am not sure about the use of parse_feature_irqs.
This function will parse interrupt info for private features which
support interrupts. For now, 3 private features, FME error, Port error
& User interrupt (for AFU), are using interrupt
I am not sure about the use of parse_feature_irqs.
If the irq parse fails, the feature fails to be created. So an old afu feature
which loaded ok in an older kernel can fail. This could surprise the user.
Below is a change that fails more gracefully. Even if there is a problem in
the parse,
DFL based FPGA devices could support interrupts for different purposes,
but current DFL framework only supports feature device enumeration with
given MMIO resources information via common DFL headers. This patch
introduces one new API dfl_fpga_enum_info_add_irq for low level bus
drivers (e.g. PCIe
DFL based FPGA devices could support interrupts for different purposes,
but current DFL framework only supports feature device enumeration with
given MMIO resources information via common DFL headers. This patch
introduces one new API dfl_fpga_enum_info_add_irq for low level bus
drivers (e.g. PCIe
6 matches
Mail list logo