Thanks for the review Ilpo, I will do a v3 with your feedback. Regards, Fernando. ________________________________________ From: Ilpo Järvinen <ilpo.jarvinen@xxxxxxxxxxxxxxx> Sent: Friday, September 15, 2023 1:18 PM To: Fernando Eckhardt Valle (FIPT) Cc: Hans de Goede; Mark Pearson; corbet@xxxxxxx; hmh@xxxxxxxxxx; markgross@xxxxxxxxxx; linux-doc@xxxxxxxxxxxxxxx; LKML; ibm-acpi-devel@xxxxxxxxxxxxxxxxxxxxx; platform-driver-x86@xxxxxxxxxxxxxxx Subject: Re: [PATCH v2] platform/x86: thinkpad_acpi: sysfs interface to auxmac On Fri, 15 Sep 2023, Fernando Eckhardt Valle wrote: > Newer Thinkpads have a feature called Mac Address Passthrough. > This patch provides a sysfs interface that userspace can use > to get this auxiliary mac address. > > Changes in v2: > - Added documentation > - All handling of the auxmac value is done in the _init function. > > Signed-off-by: Fernando Eckhardt Valle <fevalle@xxxxxx> > --- > .../admin-guide/laptops/thinkpad-acpi.rst | 20 +++++ > drivers/platform/x86/thinkpad_acpi.c | 78 +++++++++++++++++++ > 2 files changed, 98 insertions(+) > > diff --git a/Documentation/admin-guide/laptops/thinkpad-acpi.rst b/Documentation/admin-guide/laptops/thinkpad-acpi.rst > index e27a1c3f6..6207c363f 100644 > --- a/Documentation/admin-guide/laptops/thinkpad-acpi.rst > +++ b/Documentation/admin-guide/laptops/thinkpad-acpi.rst > @@ -53,6 +53,7 @@ detailed description): > - Lap mode sensor > - Setting keyboard language > - WWAN Antenna type > + - Auxmac > > A compatibility table by model and feature is maintained on the web > site, http://ibm-acpi.sf.net/. I appreciate any success or failure > @@ -1511,6 +1512,25 @@ Currently 2 antenna types are supported as mentioned below: > The property is read-only. If the platform doesn't have support the sysfs > class is not created. > > +Auxmac > +------ > + > +sysfs: auxmac > + > +Some newer Thinkpads have a feature called MAC Address Passthrough. This > +feature is implemented by the system firmware to provide a system unique MAC, > +that can override a dock or USB ethernet dongle MAC, when connected to a > +network. This property enables user-space to easily determine the MAC address > +if the feature is enabled. > + > +The values of this auxiliary MAC are: > + > + cat /sys/devices/platform/thinkpad_acpi/auxmac > + > +If the feature is disabled, the value will be 'disabled'. > + > +This property is read-only. > + > Adaptive keyboard > ----------------- > > diff --git a/drivers/platform/x86/thinkpad_acpi.c b/drivers/platform/x86/thinkpad_acpi.c > index d70c89d32..05cc3a1e2 100644 > --- a/drivers/platform/x86/thinkpad_acpi.c > +++ b/drivers/platform/x86/thinkpad_acpi.c > @@ -10785,6 +10785,79 @@ static struct ibm_struct dprc_driver_data = { > .name = "dprc", > }; > > +/* > + * Auxmac > + * > + * This auxiliary mac address is enabled in the bios through the > + * Mac Address Passthrough feature. In most cases, there are three > + * possibilities: Internal Mac, Second Mac, and disabled. > + * > + */ > + > +#define AUXMAC_LEN 12 > +#define AUXMAC_START 9 > +#define AUXMAC_STRLEN 22 > +static char auxmac[AUXMAC_LEN]; > + > +static int auxmac_init(struct ibm_init_struct *iibm) > +{ > + acpi_status status; > + struct acpi_buffer buffer = { ACPI_ALLOCATE_BUFFER, NULL }; > + union acpi_object *obj; > + > + status = acpi_evaluate_object(NULL, "\\MACA", NULL, &buffer); > + > + if (ACPI_FAILURE(status)) > + return -ENODEV; > + > + obj = (union acpi_object *)buffer.pointer; > + > + if (obj->type != ACPI_TYPE_STRING || obj->string.length != AUXMAC_STRLEN) { > + pr_info("Invalid buffer for mac addr passthrough.\n"); MAC address > + goto auxmacinvalid; > + } > + > + if (strncmp(obj->string.pointer + 0x8, "#", 1) != 0 || > + strncmp(obj->string.pointer + 0x15, "#", 1) != 0) { Why use strncmp with (..., 1)? These offsets should defines above and not use literals. > + pr_info("Invalid header for mac addr passthrough.\n"); MAC address > + goto auxmacinvalid; > + } > + > + if (strncmp(obj->string.pointer + 0x9, "XXXXXXXXXXXX", AUXMAC_LEN) == 0) Why you're not using AUXMAC_START here? It's also bit confusing that some of the offset are hex and some non-hex numbers. > + memcpy(auxmac, "disabled", 9); Don't use memcpy() for copying a string. > + else > + memcpy(auxmac, obj->string.pointer + AUXMAC_START, AUXMAC_LEN); What about the termination of auxmac? It's given > + > + kfree(obj); > + return 0; > + > +auxmacinvalid: > + kfree(obj); > + memcpy(auxmac, "unavailable", 11); Again, don't use memcpy() to copy a string. You even got it wrong here compared with the other case where you copied also the zero terminator. > + return 0; > +} > + > +static struct ibm_struct auxmac_data = { > + .name = "auxmac", > +}; > + > +static ssize_t auxmac_show(struct device *dev, > + struct device_attribute *attr, > + char *buf) > +{ > + return sysfs_emit(buf, "%s\n", auxmac); This requires proper termination for the string but you didn't ensure it above. -- i. > +} > +static DEVICE_ATTR_RO(auxmac); > + > +static struct attribute *auxmac_attributes[] = { > + &dev_attr_auxmac.attr, > + NULL > +}; > + > +static const struct attribute_group auxmac_attr_group = { > + .attrs = auxmac_attributes, > +}; > + > /* --------------------------------------------------------------------- */ > > static struct attribute *tpacpi_driver_attributes[] = { > @@ -10843,6 +10916,7 @@ static const struct attribute_group *tpacpi_groups[] = { > &proxsensor_attr_group, > &kbdlang_attr_group, > &dprc_attr_group, > + &auxmac_attr_group, > NULL, > }; > > @@ -11414,6 +11488,10 @@ static struct ibm_init_struct ibms_init[] __initdata = { > .init = tpacpi_dprc_init, > .data = &dprc_driver_data, > }, > + { > + .init = auxmac_init, > + .data = &auxmac_data, > + }, > }; > > static int __init set_ibm_param(const char *val, const struct kernel_param *kp) >
