NFC configurations on FP2 device configs (WTF)

[Roboe]

As some of you already know, I’m porting CyanogenMod for the FP2 with some other great people like @NicoM, @Arvil and @z3ntu (plus others).

While doing so, rewriting the device configs, I stumbled upon the below lines:

TARGET_USES_QCA_NFC := other

# NFC packages
ifeq ($(TARGET_USES_QCA_NFC),true)
NFC_D := true

ifeq ($(NFC_D), true)
    PRODUCT_PACKAGES += \
        libnfcD-nci \
        libnfcD_nci_jni \
        nfc_nci.msm8974 \
        NfcDNci \
        Tag \
        com.android.nfc_extras \
        com.android.nfc.helper
else
PRODUCT_PACKAGES += \
    libnfc-nci \
    libnfc_nci_jni \
    nfc_nci.msm8974 \
    NfcNci \
    Tag \
    com.android.nfc_extras
endif

# file that declares the MIFARE NFC constant
# Commands to migrate prefs from com.android.nfc3 to com.android.nfc
# NFC access control + feature files + configuration
PRODUCT_COPY_FILES += \
        frameworks/native/data/etc/com.nxp.mifare.xml:system/etc/permissions/com.nxp.mifare.xml \
        frameworks/native/data/etc/com.android.nfc_extras.xml:system/etc/permissions/com.android.nfc_extras.xml \
        frameworks/native/data/etc/android.hardware.nfc.xml:system/etc/permissions/android.hardware.nfc.xml
# Enable NFC Forum testing by temporarily changing the PRODUCT_BOOT_JARS
# line has to be in sync with build/target/product/core_base.mk
endif

Questions flow immediately:

• Why are those files configuring NFC while the FP2 has no hardware support for this?
• Could this be related to the extreme battery consumption some users suffer from?
• Are they there for some reason, or are just CodeAurora remainders? If the former, should we include this configs when porting?

Let’s see if someone from the developers team (@jftr, @dirkv) can take some look here or tell their coworkers.

[jftr]

The answer lies in the first line you post:

TARGET_USES_QCA_NFC := other

The entire block is only executed, if TARGET_USES_QCA_NFC equals true. However, in above line the variable is set to other. This means the entire block you posted is not evaluated.

Edit: To extend my answer and give some background:

AOSP source tree includes essentially everything that is necessary to use any feature of Android. With configuration files for each device, one can choose and pick what is enabled based on device hardware and capabilities. Of course, AOSP also includes support for NFC. Only for FP2 this is not enabled.

The same is also true for any features Qualcomm adds (the QCA in the variable name suggests, this is a Qualcomm feature): Qualcomm adds code for all kind of features supported by its chips. It is up to the phone manufacturer to select those that make sense for their device. In this case, TARGET_USES_QCA_NFC is disabled for obvious reasons

[jftr]

Ah, maybe I missed the point of your question. Did you rather ask why the entire block is even included in “FP2.mk”?

I don’t know about the specifics (I am working on FP1 after all), but my guess is that the makefile for FP2 was created by copying a template provided by Qualcomm and change it where necessary. The minimal change to disable NFC is to set this variable to anything not equal to true. Of course, just keeping the line TARGET_USES_QCA_NFC := other and removing the entire ifeq block would lead to the same result.

Edit: If you port to CM by starting with a template from CM and copy over everything you need from FP2.mk, then do not include the block. You probably do not even need the TARGET_USES_QCA_NFC := other, however I cannot say the latter with 100% certainty.

[Roboe]

The entire block is only executed, if TARGET_USES_QCA_NFC equals true. However, in above line the variable is set to other. This means the entire block you posted is not evaluated.

I completely overlook that conditional! Even with its content indented,

It confounded me the other value. Why not just false? Hahaha, . Semantically, other could mean “I’ll configure this on my own” (e.g. with another package on the device config plus the respective hardware subfolder). Makefile “language” has its limitations… hahaha. (Not an issue, just my little-but-growing knowledge about GNUMake)

If you port to CM by starting with a template from CM and copy over everything you need from FP2.mk, then do not include the block. You probably do not even need the TARGET_USES_QCA_NFC := other, however I cannot say the latter with 100% certainty.

Yes, we are rewriting the device configs from scratch (i.e. from the succinct base CM provides). That approach has the advantage of allow us to learn one step at a time. We’ll inform you!

Thank you for all the clarifications!

[Nigelgreenx]

Are there any plans to enable nfc hardware

[ced117]

@Nigelgreenx There is no NFC hardware in the Fairphone 2.

[TobiasF]

There is no NFC hardware in the Fairphone 2.

But there could be a new back cover in the future which enables NFC on the Fairphone 2. (Well, the NFC hardware is brought by the back cover in this case.)

[ced117]

@TobiasF Yes, true
I hope that this could happen one day.

[Roboe]

Well, the hardware is in fact modular, but Android isn’t…

I don’t know how difficult could be make Android detect hardware at runtime to enable hardware-dependent features (like NFC), but I doubt is easy and I doubt we’ll see it in the FP2.

Anyway, that’s another topic to discuss on other forum category and this one already came to its end,

[jaymanu]

make Android detect hardware at runtime

what about USB interface ?