[<prev] [next>] [<thread-prev] [thread-next>] [day] [month] [year] [list]
Message-ID: <df4a01cc-859f-4b86-aa86-58bc03fefb02@amd.com>
Date: Wed, 22 May 2024 11:50:49 -0500
From: Ben Cheatham <benjamin.cheatham@....com>
To: Zaid Alali <zaidal@...amperecomputing.com>, rafael@...nel.org,
lenb@...nel.org, james.morse@....com, tony.luck@...el.com, bp@...en8.de,
robert.moore@...el.com, Jonathan.Cameron@...wei.com,
dan.j.williams@...el.com, arnd@...db.de, Avadhut.Naik@....com,
u.kleine-koenig@...gutronix.de, john.allen@....com,
linux-acpi@...r.kernel.org, linux-kernel@...r.kernel.org,
acpica-devel@...ts.linux.dev
Subject: Re: [RFC PATCH v2 7/8] ACPI: APEI: EINJ: Enable EINJv2 error
injections
On 5/21/24 4:10 PM, Zaid Alali wrote:
> Enable the driver to inject EINJv2 type errors. The component
> array values are parsed from user_input and expected to contain
> hex values for component id and syndrome separated by space,
> and multiple components are separated by new line as follows:
>
> component_id1 component_syndrome1
> component_id2 component_syndrome2
> :
> component_id(n) component_syndrome(n)
>
> for example:
>
> $comp_arr="0x1 0x2
>> 0x1 0x4
>> 0x2 0x4"
> $cd /sys/kernel/debug/apei/einj/
> $echo "$comp_arr" > einjv2_component_array
>
I think it would be good to change this from being newline-delimited to comma-delimited instead.
So instead of your first example above it would be:
component_id1 component_syndrome1, component_id2 component_syndrome2, ...
My reasoning here is that it's less error-prone. For example, if I run your example but forget to
quote the comp_arr variable in the last line, i.e.:
$ echo $comp_arr > einjv2_component_array
I would effectively be running (at least in my stock Ubuntu 22.04 terminal):
$ echo "0x1 0x2 0x1 0x4 0x2 0x4" > einjv2_component_array
Which would result in an error for something that isn't necessarily readily apparent.
I also think keeping the input on a single line is nicer on the eyes, but that's a subjective thing
and I'd understand if you think differently.
> Signed-off-by: Zaid Alali <zaidal@...amperecomputing.com>
> ---
> drivers/acpi/apei/einj-core.c | 81 ++++++++++++++++++++++++++++++++---
> 1 file changed, 75 insertions(+), 6 deletions(-)
>
> diff --git a/drivers/acpi/apei/einj-core.c b/drivers/acpi/apei/einj-core.c
> index 2e30ebed079b..2e5c00b34a4b 100644
> --- a/drivers/acpi/apei/einj-core.c
> +++ b/drivers/acpi/apei/einj-core.c
> @@ -87,6 +87,13 @@ enum {
> SETWA_FLAGS_APICID = 1,
> SETWA_FLAGS_MEM = 2,
> SETWA_FLAGS_PCIE_SBDF = 4,
> + SETWA_FLAGS_EINJV2 = 8,
> +};
> +
> +enum {
> + EINJV2_PROCESSOR_ERROR = 0x1,
> + EINJV2_MEMORY_ERROR = 0x2,
> + EINJV2_PCIE_ERROR = 0x4,
> };
>
> /*
> @@ -111,6 +118,7 @@ static char vendor_dev[64];
> static struct debugfs_blob_wrapper einjv2_component_arr;
> static u64 component_count;
> static void *user_input;
> +static int nr_components;
> static u32 available_error_type;
> static u32 available_error_type_v2;
>
> @@ -287,8 +295,18 @@ static void *einj_get_parameter_address(void)
>
> v5param = acpi_os_map_iomem(pa_v5, sizeof(*v5param));
> if (v5param) {
> + int offset, len;
> +
> acpi5 = 1;
> check_vendor_extension(pa_v5, v5param);
> + if (available_error_type & ACPI65_EINJV2_SUPP) {
> + len = v5param->einjv2_struct.length;
> + offset = offsetof(struct einjv2_extension_struct, component_arr);
> + nr_components = (len - offset) / 32;
> + acpi_os_unmap_iomem(v5param, sizeof(*v5param));
> + v5param = acpi_os_map_iomem(pa_v5, sizeof(*v5param) + (
> + (nr_components) * sizeof(struct syndrome_array)));
> + }
> return v5param;
> }
> }
> @@ -494,10 +512,52 @@ static int __einj_error_inject(u32 type, u32 flags, u64 param1, u64 param2,
> v5param->flags = vendor_flags;
> } else if (flags) {
> v5param->flags = flags;
> - v5param->memory_address = param1;
> - v5param->memory_address_range = param2;
> - v5param->apicid = param3;
> - v5param->pcie_sbdf = param4;
> + if (flags & SETWA_FLAGS_MEM) {
> + v5param->memory_address = param1;
> + v5param->memory_address_range = param2;
> + }
I don't think you need this if statement since the values will be ignored
when the SETWA_FLAGS_MEM bit isn't set anyway as per the spec.
> + if (flags & SETWA_FLAGS_EINJV2) {
> + int count = 0, bytes_read, pos = 0;
> + unsigned int comp, synd;
> + struct syndrome_array *component_arr;
> +
> + if (component_count > nr_components)
> + goto err_out;
> +
> + v5param->einjv2_struct.component_arr_count = component_count;
> + component_arr = v5param->einjv2_struct.component_arr;
> +
> + while (sscanf(user_input+pos, "%x %x\n%n", &comp, &synd,
> + &bytes_read) == 2) {
> + count++;
> + pos += bytes_read;
> + if (count > component_count)
> + goto err_out;
> +
> + switch (type) {
> + case EINJV2_PROCESSOR_ERROR:
> + component_arr[count-1].comp_id.acpi_id = comp;
> + component_arr[count-1].comp_synd.proc_synd = synd;
> + break;
> + case EINJV2_MEMORY_ERROR:
> + component_arr[count-1].comp_id.device_id = comp;
> + component_arr[count-1].comp_synd.mem_synd = synd;
> + break;
> + case EINJV2_PCIE_ERROR:
> + component_arr[count-1].comp_id.pcie_sbdf = comp;
> + component_arr[count-1].comp_synd.pcie_synd = synd;
> + break;
> + }
> + }
> + if (count != component_count)
> + goto err_out;
Nitpick here, but you could use count directly instead of count-1 when indexing component_arr[]
if you move the count++ to the bottom of the loop and change the above if to:
if (count != component_count - 1)
goto err_out;
or use count + 1 instead of component_count - 1.
> +
> + /* clear buffer after user input for next injection */
> + memset(user_input, 0, COMP_ARR_SIZE);
> + } else {
> + v5param->apicid = param3;
> + v5param->pcie_sbdf = param4;
> + }
> } else {
> switch (type) {
> case ACPI_EINJ_PROCESSOR_CORRECTABLE:
> @@ -570,6 +630,9 @@ static int __einj_error_inject(u32 type, u32 flags, u64 param1, u64 param2,
> rc = apei_exec_run_optional(&ctx, ACPI_EINJ_END_OPERATION);
>
> return rc;
> +err_out:
> + memset(user_input, 0, COMP_ARR_SIZE);
> + return -EINVAL;
> }
>
> /* Inject the specified hardware error */
> @@ -581,9 +644,14 @@ int einj_error_inject(u32 type, u32 flags, u64 param1, u64 param2, u64 param3,
>
> /* If user manually set "flags", make sure it is legal */
> if (flags && (flags &
> - ~(SETWA_FLAGS_APICID|SETWA_FLAGS_MEM|SETWA_FLAGS_PCIE_SBDF)))
> + ~(SETWA_FLAGS_APICID|SETWA_FLAGS_MEM|SETWA_FLAGS_PCIE_SBDF|SETWA_FLAGS_EINJV2)))
> return -EINVAL;
>
> + /*check if type is a valid EINJv2 error type*/
> + if (flags & SETWA_FLAGS_EINJV2) {
> + if (!(type & available_error_type_v2))
> + return -EINVAL;
> + }
> /*
> * We need extra sanity checks for memory errors.
> * Other types leap directly to injection.
> @@ -915,7 +983,8 @@ static void __exit einj_remove(struct platform_device *pdev)
> sizeof(struct set_error_type_with_address) :
> sizeof(struct einj_parameter);
>
> - acpi_os_unmap_iomem(einj_param, size);
> + acpi_os_unmap_iomem(einj_param,
> + size + (nr_components * sizeof(struct syndrome_array)));
> if (vendor_errors.size)
> acpi_os_unmap_memory(vendor_errors.data, vendor_errors.size);
> }
Powered by blists - more mailing lists