RPMB and GPT Listeners - Merge Request #14

quic-bhaskarv · 2025-11-17T02:57:30Z

This change implements RPMB and GPT listeners, please review.

harshaldev27 · 2025-11-23T06:33:37Z

listeners/libgptservice/CMakeLists.txt

+# Install the shared library
+install(TARGETS gptservice
+    LIBRARY DESTINATION ${CMAKE_INSTALL_LIBDIR}
+    ARCHIVE DESTINATION ${CMAKE_INSTALL_LIBDIR}


Why are we installing ARCHIVE and RUNTIME when this is just a shared library?

I installed these to compile and test as independent package (as an experiment). I have removed it.

harshaldev27 · 2025-11-23T06:35:23Z

listeners/libgptservice/CMakeLists.txt

+target_link_libraries(gptservice
+	PRIVATE minkadaptor
+	PRIVATE z
+	PRIVATE uuid


Have you attempted a standalone build of this package with your changes? Since libz and libuuid aren't available inside this repo, my suspicion is unless you explicitly try to find them first with find_package() they should give a build-time error.

I have compiled standalone build, and tested. however, the libraries are already compiled and haven't faced issues in compilation. Fixed it in latest change.

qtee_supplicant/src/listener_mngr.c

harshaldev27 · 2025-11-23T07:03:18Z

listeners/libgptservice/CMakeLists.txt

+)
+
+# Install header files
+install(FILES


I don't think a listener library is supposed to install and export headers. The only client for a listener is qtee_supplicant, which only calls init() and deinit() functions exported by the library. There is also no need for headers because the listener and qtee_supplicant are always separately built, with not build-time dependency on each other.

Done, removed.

harshaldev27 · 2025-11-23T07:04:35Z

listeners/libgptservice/gpt_test_client

Why are you pushing a prebuilt binary here? This is not allowed. Build all your test clients by adding their sources here.

harshaldev27 · 2025-11-23T07:08:38Z

listeners/libgptservice/gpt_test_client.c

+	print_guid(req.guid);
+	printf("\n");
+
+	/* Simulate response (in real implementation, this would come from TEE) */


This is not the proper way to ensure testing for this listener library. This is just a test for the sake of adding a test, I cannot allow it. Please create a Trusted Application which makes a call into the GPT listener and push the binary for that Trusted Application in this repository, so that developers with access to a QC board can actually validate that the GPT listener in this repository is being properly invoked by QTEE.

removed, it was tested internally. It is not required now.

harshaldev27 · 2025-11-23T07:10:50Z

listeners/libgptservice/CMakeLists.txt

+
+target_compile_definitions(gptservice
+	PRIVATE
+		-DGPT_LISTENER


I don't think you are using this flag in the GPT listener, it is unnecessary.

harshaldev27 · 2025-11-23T07:23:44Z

listeners/libgptservice/CMakeLists.txt

+)
+
+# GPT Test Client (optional)
+add_executable(gpt_test_client


You shouldn't add a new test client for GPT, add your test case to smcinvoke_client itself.

listeners/libgptservice/README.md

harshaldev27 · 2025-11-23T07:26:08Z

listeners/libgptservice/README.md

+### Dependencies
+- zlib (for CRC32 calculations)
+- uuid (for GUID operations)
+- Linux block device support


What kind of block device is expected here? Please provide more information. For example, for QCOMTEE we need the /dev/tee0 node, accessed by the QCOMTEE library. We have provided information that the TEE subsystem exposes this node.

added block device details.

harshaldev27 · 2025-11-23T07:27:38Z

Please ensure that the name of the Pull Request is a concise title which highlights the individual commits in the Pull Request. Just saying "New listeners" makes it difficult to find the Pull Request later which added support for GPT and RPMB listener.

harshaldev27

I don't think you need a separate commit for this. Please rebase your branch ontop of minkipc tip.

harshaldev27 · 2026-01-02T08:59:19Z

listeners/libgptservice/gpt_logging.c

@@ -0,0 +1,87 @@
+// Copyright (c) 2024 Qualcomm Technologies, Inc.


Copyright update please.

harshaldev27 · 2026-01-02T08:59:53Z

listeners/libgptservice/gpt_logging.h

@@ -0,0 +1,68 @@
+// Copyright (c) 2024 Qualcomm Technologies, Inc.


Copyright update required.

Okay, will remove copyright.

Don't remove, update it to yearless as in gpt_msg.h

listeners/libgptservice/gpt_msg.h

harshaldev27 · 2026-01-02T09:02:23Z

listeners/libgptservice/gpt_msg.h

+#define MSGE(fmt, ...) GPT_LOG_ERROR(fmt, ##__VA_ARGS__)
+#define MSGD(fmt, ...) GPT_LOG_DEBUG(fmt, ##__VA_ARGS__)
+
+/* Fixed. Don't increase the size of TZ_CM_MAX_NAME_LEN */


Hmm.. I see many listeners are using these macros, maybe we should think about moving them to a separate common file eventually. For now it is good.

Yes, logging format can be same and we can modify/use for all listeners.

listeners/libgptservice/gpt_msg.h

harshaldev27

This is not the right way @quic-bhaskarv . Please take your original 2 commits for RPMB and GPT. Rebase your fork onto tip of minkipc repo and reapply those two commits. Then force push to your minkipc fork. Creating new 'rebase' and 'merge' commits in this PR is not the way forward.

harshaldev27

I need to understand how are you testing this listener? Are you using some TA? Could you provide steps you are using as well as a small log snippet of success case?

harshaldev27 · 2026-01-02T09:09:00Z

listeners/libgptservice/gpt_service.c

+
+/* GPT constants and definitions from original service */
+#define GPT_LSTNR_VERSION_1		1
+#define GPT_LSTNR_VERSION_2		2


These two macros seem a bit redundant, why not just #define GPT_LSTNR_VERSION 2 on line 46?

we need to support 2 version, please check the code in smci_gpt_handle_init_req for details.

listeners/libgptservice/gpt_msg.h

harshaldev27 · 2026-01-02T09:24:02Z

listeners/libgptservice/gpt_service.c

+} gpt_init_info_t;
+
+struct gpt_stats {
+	int		init_done;


Please avoid adding unnecessary spaces between member type and name. Just int init_done is fine.

Same for below as well.

this is for readability purpose, removed the tabs.

harshaldev27 · 2026-01-02T09:27:22Z

listeners/libgptservice/gpt_logging.h

@@ -0,0 +1,68 @@
+// Copyright (c) 2024 Qualcomm Technologies, Inc.


Don't remove, update it to yearless as in gpt_msg.h

harshaldev27 · 2026-01-02T09:28:04Z

listeners/libgptservice/gpt_service.c

+} __attribute__ ((packed));
+
+struct disk {
+	char			dev_name[FNAME_SZ];


Avoid unnecessary spaces in struct members.

harshaldev27 · 2026-01-02T09:41:31Z

listeners/libgptservice/gpt_service.c

+			if (dmaBufLen >= sizeof(tz_gpt_rw_res_t)) {
+				smci_gpt_handle_rw_req(req, rsp);
+			} else {
+				MSGE("Invalid input.");


Let's print something more concrete like, "Response buffer length is too small!" to that we know whether this is a transport issue or a use-case issue.

harshaldev27 · 2026-01-02T09:43:07Z

listeners/libgptservice/gpt_service.c

+
+	MSGD("GPT_SERVICE Dispatch ends! ");
+
+	free(req);


This is an unnecessary allocation let's try to avoid it unless you have a very valid reason.

harshaldev27 · 2026-01-02T09:47:12Z

listeners/libgptservice/gpt_service.c

+		init_resp.ctx_id = -1;
+		init_resp.status = GPT_LSTNR_VER_NOT_SUPPORTED;
+
+		memmove(rsp,  (void*)&init_resp,


I think this doesn't require wrap around.

harshaldev27 · 2026-01-02T09:49:18Z

listeners/libgptservice/gpt_service.c

+/*
+ * Main dispatch function for GPT service - matches original exactly
+ */
+int smci_dispatch(void *dmabuff, size_t dmaBufLen)


You should avoid mixing snake case and camelCase variable names. I prefer you use snake_case dma_buf_len since the rest of this file does that.

harshaldev27 · 2026-01-02T09:55:48Z

listeners/libgptservice/gpt_service.c

+	}
+}
+
+static void smci_gpt_handle_rw_req(void *req, void *rsp)


Could we benefit from providing a brief comment explaining these functions? Other listeners are doing it.

listeners/librpmbservice/CMakeLists.txt

listeners/librpmbservice/README.md

listeners/librpmbservice/comdef.h

listeners/librpmbservice/rpmb.c

listeners/librpmbservice/comdef.h

harshaldev27 · 2026-01-02T10:13:32Z

listeners/librpmbservice/rpmb.c

+
+static struct rpmb_wake_lock
+{
+	int       lock_fd;


Avoid unnecessary space between datatype and variable name.

listeners/librpmbservice/rpmb.h

listeners/librpmbservice/rpmb_core.c

listeners/librpmbservice/rpmb_service.c

listeners/librpmbservice/rpmb_stubs.c

This change introduces GPT listener service to support secure GPT requests from QTEE. Change-Id: I6d6b5f5c2701735253e72d6284ce41234142ddbb Signed-off-by: Bhaskar Valaboju <quic_bhaskarv@quicinc.com>

This change introduces RPMB listener service to support secure RPMB requests from QTEE. Change-Id: Ic46489cc366956a946b7e02512cd8f5432af0133 Signed-off-by: Bhaskar Valaboju <quic_bhaskarv@quicinc.com>

harshaldev27 reviewed Nov 23, 2025

View reviewed changes

quic-bhaskarv force-pushed the new_listeners branch from bf4a3e1 to 02fdb20 Compare December 30, 2025 06:21

quic-bhaskarv changed the title ~~New listeners~~ RPMB and GPT Listeners - Merge Request Dec 30, 2025

harshaldev27 reviewed Jan 2, 2026

View reviewed changes

harshaldev27 requested changes Jan 2, 2026

View reviewed changes

quic-bhaskarv force-pushed the new_listeners branch 2 times, most recently from 5c2694b to da25817 Compare January 2, 2026 09:29

harshaldev27 requested changes Jan 2, 2026

View reviewed changes

harshaldev27 reviewed Jan 2, 2026

View reviewed changes

listeners/librpmbservice/CMakeLists.txt Outdated Show resolved Hide resolved

harshaldev27 requested changes Jan 2, 2026

View reviewed changes

quic-bhaskarv added 2 commits January 21, 2026 15:07

Add support for GPT listener

4143a8e

This change introduces GPT listener service to support secure GPT requests from QTEE. Change-Id: I6d6b5f5c2701735253e72d6284ce41234142ddbb Signed-off-by: Bhaskar Valaboju <quic_bhaskarv@quicinc.com>

Add support for RPMB listener

667875c

This change introduces RPMB listener service to support secure RPMB requests from QTEE. Change-Id: Ic46489cc366956a946b7e02512cd8f5432af0133 Signed-off-by: Bhaskar Valaboju <quic_bhaskarv@quicinc.com>

quic-bhaskarv force-pushed the new_listeners branch from 7691f18 to 667875c Compare January 21, 2026 11:42

		@@ -0,0 +1,87 @@
		// Copyright (c) 2024 Qualcomm Technologies, Inc.

		@@ -0,0 +1,68 @@
		// Copyright (c) 2024 Qualcomm Technologies, Inc.

RPMB and GPT Listeners - Merge Request #14

Are you sure you want to change the base?

RPMB and GPT Listeners - Merge Request #14

Conversation

quic-bhaskarv commented Nov 17, 2025

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

harshaldev27 commented Nov 23, 2025

Uh oh!

harshaldev27 left a comment

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Uh oh!

harshaldev27 left a comment

Choose a reason for hiding this comment

Uh oh!

harshaldev27 left a comment

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!