fix(cloud_proxy): harden parsing and async proxy error handling#665
fix(cloud_proxy): harden parsing and async proxy error handling#665
Conversation
jkralik
commented
Apr 9, 2026
jkralik
commented
- Prevent overflow when building discovered resource path by validating combined UDN and URI length before copy.
- Validate anchor and URL formats before extracting UDN/local URI, and reject malformed inputs safely.
- Add bounds checks for delayed response payload copies based on separate-response buffer limits.
- Add null checks for delayed response allocations and handle dispatch/init failures for GET/POST/DELETE with explicit error responses and cleanup.
- Fix delayed response lifetime for DELETE callback to avoid leaks.
- Harden d2dserverlist DI handling with length validation, exact-length matching, and safe copy with explicit null termination.
- Avoid incorrect rediscovery on already-known DI by tracking newly added entries before issuing discovery.
- Replace unsafe deviceuuid query concatenation with bounded formatting and overflow check.
- Prevent overflow when building discovered resource path by validating combined UDN and URI length before copy. - Validate anchor and URL formats before extracting UDN/local URI, and reject malformed inputs safely. - Add bounds checks for delayed response payload copies based on separate-response buffer limits. - Add null checks for delayed response allocations and handle dispatch/init failures for GET/POST/DELETE with explicit error responses and cleanup. - Fix delayed response lifetime for DELETE callback to avoid leaks. - Harden d2dserverlist DI handling with length validation, exact-length matching, and safe copy with explicit null termination. - Avoid incorrect rediscovery on already-known DI by tracking newly added entries before issuing discovery. - Replace unsafe deviceuuid query concatenation with bounded formatting and overflow check.
WalkthroughHardened cloud proxy parsing, D2D device handling, separate-response buffering, proxied resource operations, and discovery flows with input validation, explicit length handling, safer copying, and added error-return signaling; small test headers/enum comparison fixes added. Changes
Estimated code review effort🎯 4 (Complex) | ⏱️ ~60 minutes 🚥 Pre-merge checks | ✅ 3✅ Passed checks (3 passed)
✏️ Tip: You can configure your own custom pre-merge checks in the settings. ✨ Finishing Touches📝 Generate docstrings
🧪 Generate unit tests (beta)
Thanks for using CodeRabbit! It's free for OSS, and your support helps us grow. If you like it, consider giving us a shout-out. Comment |
|
🎉 Thank you for your code contribution! To guarantee the change/addition is conformant to the OCF Specification, we would like to ask you to execute OCF Conformance Testing of your change ☝️ when your work is ready to be reviewed. ℹ️ To verify your latest change (aefeb90), label this PR with |
![coderabbitai[bot]](https://avatars.githubusercontent.com/in/347564?s=60&v=4){kind=small}
There was a problem hiding this comment.
Actionable comments posted: 1
🧹 Nitpick comments (1)
apps/cloud_proxy.c (1)
1179-1194: Good error handling; consider removing unnecessary(int)cast.The NULL check and payload size validation are excellent additions. However, line 1189 casts
_payload_lentointunnecessarily—memcpyexpectssize_t. While the bounds check above prevents practical issues, the cast is a code smell.Suggested cleanup
- memcpy(delay_response->buffer, data->_payload, (int)data->_payload_len); + memcpy(delay_response->buffer, data->_payload, data->_payload_len);🤖 Prompt for AI Agents
Verify each finding against the current code and only fix it if needed. In `@apps/cloud_proxy.c` around lines 1179 - 1194, The memcpy call currently casts data->_payload_len to (int) which is unnecessary and a code smell; update the memcpy invocation in the block handling delay_response to pass data->_payload_len directly (or explicitly cast to size_t if the compiler complains) as the third argument, e.g., memcpy(delay_response->buffer, data->_payload, data->_payload_len), keeping the surrounding checks (delay_response NULL, payload size vs get_separate_response_buffer_size(), oc_send_separate_response, and free(delay_response)) unchanged and referencing memcpy, data->_payload_len, delay_response, and oc_send_separate_response to locate the code.
🤖 Prompt for all review comments with AI agents
Verify each finding against the current code and only fix it if needed.
Inline comments:
In `@apps/cloud_proxy.c`:
- Around line 365-371: The extraction code in apps/cloud_proxy.c that computes
copy_len from anchor_len and prefix_len and writes into udn (using variables
copy_len, udn_size, memcpy, and udn[copy_len] = '\0') currently truncates when
copy_len >= udn_size and still returns true; change the logic so that if
copy_len >= udn_size (i.e., truncation would happen) the function returns false
instead of truncating and returning true, avoiding writing a partial UDN—ensure
no buffer overflow by only copying when copy_len < udn_size and return true on
success, false on truncation or other failure.
---
Nitpick comments:
In `@apps/cloud_proxy.c`:
- Around line 1179-1194: The memcpy call currently casts data->_payload_len to
(int) which is unnecessary and a code smell; update the memcpy invocation in the
block handling delay_response to pass data->_payload_len directly (or explicitly
cast to size_t if the compiler complains) as the third argument, e.g.,
memcpy(delay_response->buffer, data->_payload, data->_payload_len), keeping the
surrounding checks (delay_response NULL, payload size vs
get_separate_response_buffer_size(), oc_send_separate_response, and
free(delay_response)) unchanged and referencing memcpy, data->_payload_len,
delay_response, and oc_send_separate_response to locate the code.
🪄 Autofix (Beta)
Fix all unresolved CodeRabbit comments on this PR:
- Push a commit to this branch (recommended)
- Create a new PR with the fixes
| size_t copy_len = anchor_len - prefix_len; | ||
| if (copy_len >= udn_size) { | ||
| copy_len = udn_size - 1; | ||
| } | ||
| memcpy(udn, anchor + prefix_len, copy_len); | ||
| udn[copy_len] = '\0'; | ||
| return true; |
There was a problem hiding this comment.
Silent truncation returns true, which may mislead callers.
If the UDN portion of the anchor exceeds udn_size, the code truncates it but still returns true. This could lead to using an invalid/partial UDN that won't match any actual device, causing silent failures downstream.
Consider returning false when truncation occurs, so callers know the extraction failed:
Suggested fix
size_t copy_len = anchor_len - prefix_len;
if (copy_len >= udn_size) {
- copy_len = udn_size - 1;
+ udn[0] = '\0';
+ return false;
}
memcpy(udn, anchor + prefix_len, copy_len);
udn[copy_len] = '\0';
return true;📝 Committable suggestion
‼️ IMPORTANT
Carefully review the code before committing. Ensure that it accurately replaces the highlighted code, contains no missing lines, and has no issues with indentation. Thoroughly test & benchmark the code to ensure it meets the requirements.
| size_t copy_len = anchor_len - prefix_len; | |
| if (copy_len >= udn_size) { | |
| copy_len = udn_size - 1; | |
| } | |
| memcpy(udn, anchor + prefix_len, copy_len); | |
| udn[copy_len] = '\0'; | |
| return true; | |
| size_t copy_len = anchor_len - prefix_len; | |
| if (copy_len >= udn_size) { | |
| udn[0] = '\0'; | |
| return false; | |
| } | |
| memcpy(udn, anchor + prefix_len, copy_len); | |
| udn[copy_len] = '\0'; | |
| return true; |
🤖 Prompt for AI Agents
Verify each finding against the current code and only fix it if needed.
In `@apps/cloud_proxy.c` around lines 365 - 371, The extraction code in
apps/cloud_proxy.c that computes copy_len from anchor_len and prefix_len and
writes into udn (using variables copy_len, udn_size, memcpy, and udn[copy_len] =
'\0') currently truncates when copy_len >= udn_size and still returns true;
change the logic so that if copy_len >= udn_size (i.e., truncation would happen)
the function returns false instead of truncating and returning true, avoiding
writing a partial UDN—ensure no buffer overflow by only copying when copy_len <
udn_size and return true on success, false on truncation or other failure.
There was a problem hiding this comment.
Pull request overview
This PR hardens cloud_proxy parsing and asynchronous proxying paths to reduce overflow risk and improve error handling/cleanup across discovery and GET/POST/DELETE proxy flows.
Changes:
- Strengthened URL/anchor parsing (null/format checks, bounded copies) and safer DI handling in
/d2dserverlist. - Added bounds checks and allocation/dispatch failure handling for separate responses in proxied GET/POST/DELETE flows.
- Prevented path/query construction overflows in discovery and
deviceuuid=query building.
💡 Add Copilot custom instructions for smarter, more guided reviews. Learn how to get started.
| size_t copy_len = anchor_len - prefix_len; | ||
| if (copy_len >= udn_size) { | ||
| copy_len = udn_size - 1; | ||
| } | ||
| memcpy(udn, anchor + prefix_len, copy_len); | ||
| udn[copy_len] = '\0'; | ||
| return true; |
There was a problem hiding this comment.
anchor_to_udn silently truncates the extracted UDN when the anchor value is longer than the destination buffer, but still returns true. Using a truncated device identifier can lead to incorrect device matching (and makes malformed/oversized anchors look valid). Prefer returning false (and leaving udn empty) when the UDN does not fully fit in udn_size.
| url_to_local_url(const char *url, size_t url_size, char *local_url, | ||
| size_t local_url_size) | ||
| { |
There was a problem hiding this comment.
url_to_local_url now expects url_size to be the URL string length (callers pass url_len), but the surrounding doc comment still describes url_size as the input URL buffer size. This mismatch makes the API easy to misuse. Please update the comment (and/or rename the parameter) to reflect that url_size is the string length.
There was a problem hiding this comment.
@copilot apply changes based on this feedback
There was a problem hiding this comment.
Fixed in ab3bd74. Renamed url_size to url_len in the function signature and updated the doc comment to clarify it's the string length (not including the null-terminator), matching how all callers pass url_len.
| size_t max_payload = get_separate_response_buffer_size(); | ||
| if (data->_payload_len > max_payload) { | ||
| OC_PRINTF("ERROR: payload too large for separate response buffer\n"); | ||
| oc_send_separate_response(delay_response, OC_STATUS_INTERNAL_SERVER_ERROR); |
There was a problem hiding this comment.
When the local response payload exceeds the separate-response buffer, this sends OC_STATUS_INTERNAL_SERVER_ERROR. A more accurate status here is OC_STATUS_REQUEST_ENTITY_TOO_LARGE so the client can distinguish payload-size issues from server faults.
| oc_send_separate_response(delay_response, OC_STATUS_INTERNAL_SERVER_ERROR); | |
| oc_send_separate_response(delay_response, | |
| OC_STATUS_REQUEST_ENTITY_TOO_LARGE); |
| size_t max_payload = get_separate_response_buffer_size(); | ||
| if (data->_payload_len > max_payload) { | ||
| OC_PRINTF("ERROR: payload too large for separate response buffer\n"); | ||
| oc_send_separate_response(delay_response, OC_STATUS_INTERNAL_SERVER_ERROR); |
There was a problem hiding this comment.
When the local response payload exceeds the separate-response buffer, this sends OC_STATUS_INTERNAL_SERVER_ERROR. A more accurate status here is OC_STATUS_REQUEST_ENTITY_TOO_LARGE so the client can distinguish payload-size issues from server faults.
| oc_send_separate_response(delay_response, OC_STATUS_INTERNAL_SERVER_ERROR); | |
| oc_send_separate_response(delay_response, | |
| OC_STATUS_REQUEST_ENTITY_TOO_LARGE); |
| size_t max_payload = get_separate_response_buffer_size(); | ||
| if (data->_payload_len > max_payload) { | ||
| OC_PRINTF("ERROR: payload too large for separate response buffer\n"); | ||
| oc_send_separate_response(delay_response, OC_STATUS_INTERNAL_SERVER_ERROR); |
There was a problem hiding this comment.
When the local response payload exceeds the separate-response buffer, this sends OC_STATUS_INTERNAL_SERVER_ERROR. A more accurate status here is OC_STATUS_REQUEST_ENTITY_TOO_LARGE so the client can distinguish payload-size issues from server faults.
| oc_send_separate_response(delay_response, OC_STATUS_INTERNAL_SERVER_ERROR); | |
| oc_send_separate_response(delay_response, | |
| OC_STATUS_REQUEST_ENTITY_TOO_LARGE); |
| return; | ||
| } | ||
| char local_url[MAX_URI_LENGTH * 2] = { 0 }; | ||
| if (!url_to_local_url(url, ARRAY_SIZE(url), local_url, | ||
| if (!url_to_local_url(url, url_len, local_url, | ||
| ARRAY_SIZE(local_url))) { | ||
| OC_PRINTF("ERROR: Could not extract local url from url"); | ||
| return; |
There was a problem hiding this comment.
On URL parsing failures (unable to extract UDN/local URL), the handler returns without sending any response. That can leave the cloud request hanging until timeout. Please send an explicit error response (eg. OC_STATUS_BAD_REQUEST / OC_STATUS_BAD_OPTION) before returning so malformed proxy URLs are rejected cleanly.
| return; | ||
| } | ||
| char local_url[MAX_URI_LENGTH * 2] = { 0 }; | ||
| if (!url_to_local_url(url, ARRAY_SIZE(url), local_url, | ||
| if (!url_to_local_url(url, url_len, local_url, | ||
| ARRAY_SIZE(local_url))) { | ||
| OC_PRINTF("ERROR: Could not extract local url from url"); | ||
| return; |
There was a problem hiding this comment.
On URL parsing failures (unable to extract UDN/local URL), the handler returns without sending any response. That can leave the cloud request hanging until timeout. Please send an explicit error response (eg. OC_STATUS_BAD_REQUEST / OC_STATUS_BAD_OPTION) before returning so malformed proxy URLs are rejected cleanly.
| return; | ||
| } | ||
| char local_url[MAX_URI_LENGTH * 2] = { 0 }; | ||
| if (!url_to_local_url(url, ARRAY_SIZE(url), local_url, | ||
| if (!url_to_local_url(url, url_len, local_url, | ||
| ARRAY_SIZE(local_url))) { | ||
| OC_PRINTF("ERROR: Could not extract local url from url"); | ||
| return; |
There was a problem hiding this comment.
On URL parsing failures (unable to extract UDN/local URL), the handler returns without sending any response. That can leave the cloud request hanging until timeout. Please send an explicit error response (eg. OC_STATUS_BAD_REQUEST / OC_STATUS_BAD_OPTION) before returning so malformed proxy URLs are rejected cleanly.
- update enumstest to compare enum bounds using the underlying enum type with unsigned conversion, avoiding signedness and range issues - add missing algorithm include in discovery test header - apply formatting-only cleanup in cloud_proxy for consistency (no functional changes)
There was a problem hiding this comment.
Actionable comments posted: 3
Caution
Some comments are outside the diff and can’t be posted inline due to platform limitations.
⚠️ Outside diff range comments (1)
apps/cloud_proxy.c (1)
1212-1240:⚠️ Potential issue | 🟠 MajorSend an explicit error before returning from these validation branches.
These handlers still
returnon malformed proxy URIs or oversized queries before a separate response is indicated. In those cases the cloud caller gets no 4xx/5xx at all, just a hung request.Suggested fix pattern
if (url_len >= ARRAY_SIZE(url)) { OC_PRINTF("ERROR: URI buffer too small"); + oc_send_response(request, OC_STATUS_INTERNAL_SERVER_ERROR); return; } ... if (!url_to_udn(url, local_udn, ARRAY_SIZE(local_udn))) { OC_PRINTF("ERROR: Could not extract udn from url"); + oc_send_response(request, OC_STATUS_BAD_REQUEST); return; } ... if (request->query_len >= ARRAY_SIZE(query_as_string)) { OC_PRINTF("ERROR: Query buffer too small"); + oc_send_response(request, OC_STATUS_BAD_REQUEST); return; }Apply the same pattern in
get_resource,post_resource, anddelete_resource.Also applies to: 1322-1350, 1451-1479
🤖 Prompt for AI Agents
Verify each finding against the current code and only fix it if needed. In `@apps/cloud_proxy.c` around lines 1212 - 1240, The validation branches in get_resource (and similarly in post_resource and delete_resource) currently just return on errors (e.g., when url_len >= ARRAY_SIZE(url), url_to_udn(...) fails, url_to_local_url(...) fails, or request->query_len exceeds buffer) and leave the caller hanging; before each early return send an explicit HTTP error response using the same response path used elsewhere in these handlers (i.e., call the module's normal error/response helper used by get_resource/post_resource/delete_resource to send a 4xx/5xx with a descriptive message) so the cloud caller receives a proper error code instead of a hung request.
♻️ Duplicate comments (1)
apps/cloud_proxy.c (1)
365-371:⚠️ Potential issue | 🟡 MinorReject oversized anchors instead of truncating them.
Line 367 still truncates an oversized anchor and Line 371 returns
true. That feeds a partial UDN into discovery instead of rejecting malformed input.🤖 Prompt for AI Agents
Verify each finding against the current code and only fix it if needed. In `@apps/cloud_proxy.c` around lines 365 - 371, Instead of truncating oversized anchors, validate the computed copy length (copy_len = anchor_len - prefix_len) against udn_size and reject by returning false before copying; specifically, in the block that checks copy_len >= udn_size (using variables copy_len, anchor_len, prefix_len, udn_size, udn), replace the truncation logic with an early return false when copy_len > udn_size - 1 (or >= udn_size), and only perform memcpy(udn, anchor + prefix_len, copy_len) and null-terminate udn when the size fits.
🤖 Prompt for all review comments with AI agents
Verify each finding against the current code and only fix it if needed.
Inline comments:
In `@apps/cloud_proxy.c`:
- Line 1733: Run clang-format over the reported hunk to fix spacing of the
trailing comment; specifically reformat the line containing the closing brace
and the comment "/* if loop */" so it matches project style (apply clang-format
to the block containing the brace and comment and stage the updated file
apps/cloud_proxy.c). Ensure only formatting changes are committed.
- Around line 717-723: The remove_di() loop currently only clears
g_d2dserverlist_d2dserverlist[i].di, leaking the corresponding
discovered_server[i] endpoint state; update remove_di() so when a DI match is
found you also free or reset the discovered_server entry (e.g., free
discovered_server[i].endpoint or call its cleanup/reset function, zero its
struct, and reset any allocation flags) before returning true, ensuring
discovered_server[i] does not retain stale allocated memory or state for reused
slots.
- Around line 742-745: The loop that searches app resources (using
oc_ri_get_app_resources and in unregister_resources) wrongly compares di to the
resource URI at byte 0; fix it to match registered proxy URIs of the form
"/<di>/...". Specifically, when iterating oc_resource_t *res and comparing
oc_string(res->uri) to di, compare starting at the character after the leading
'/' (e.g. compare oc_string(res->uri) + 1 to di for di_len bytes) and also
require that the character immediately after the matched di in the URI is '/'
(to ensure a "/<di>/" prefix), and apply the same change to the other loop
mentioned (lines 766-771).
---
Outside diff comments:
In `@apps/cloud_proxy.c`:
- Around line 1212-1240: The validation branches in get_resource (and similarly
in post_resource and delete_resource) currently just return on errors (e.g.,
when url_len >= ARRAY_SIZE(url), url_to_udn(...) fails, url_to_local_url(...)
fails, or request->query_len exceeds buffer) and leave the caller hanging;
before each early return send an explicit HTTP error response using the same
response path used elsewhere in these handlers (i.e., call the module's normal
error/response helper used by get_resource/post_resource/delete_resource to send
a 4xx/5xx with a descriptive message) so the cloud caller receives a proper
error code instead of a hung request.
---
Duplicate comments:
In `@apps/cloud_proxy.c`:
- Around line 365-371: Instead of truncating oversized anchors, validate the
computed copy length (copy_len = anchor_len - prefix_len) against udn_size and
reject by returning false before copying; specifically, in the block that checks
copy_len >= udn_size (using variables copy_len, anchor_len, prefix_len,
udn_size, udn), replace the truncation logic with an early return false when
copy_len > udn_size - 1 (or >= udn_size), and only perform memcpy(udn, anchor +
prefix_len, copy_len) and null-terminate udn when the size fits.
🪄 Autofix (Beta)
Fix all unresolved CodeRabbit comments on this PR:
- Push a commit to this branch (recommended)
- Create a new PR with the fixes
ℹ️ Review info
⚙️ Run configuration
Configuration used: Repository UI
Review profile: CHILL
Plan: Pro
Run ID: b154a773-dadb-4aaf-bb8b-a89570c635a4
📒 Files selected for processing (3)
api/unittest/discovery/discovery.hapi/unittest/enumstest.cppapps/cloud_proxy.c
✅ Files skipped from review due to trivial changes (1)
- api/unittest/discovery/discovery.h
| for (int i = 0; i < MAX_ARRAY; i++) { | ||
| OC_PRINTF(" %s %.*s ", g_d2dserverlist_d2dserverlist[i].di, len, di); | ||
| if (strncmp(g_d2dserverlist_d2dserverlist[i].di, di, len) == 0) { | ||
| size_t stored_len = strlen(g_d2dserverlist_d2dserverlist[i].di); | ||
| if (stored_len == (size_t)len && | ||
| strncmp(g_d2dserverlist_d2dserverlist[i].di, di, (size_t)len) == 0) { | ||
| strcpy(g_d2dserverlist_d2dserverlist[i].di, ""); | ||
| return true; |
There was a problem hiding this comment.
Free the endpoint slot when removing a DI.
remove_di() only clears g_d2dserverlist_d2dserverlist[i].di. The endpoint copy in discovered_server[i] stays allocated, so add/remove cycles leak memory and reuse a slot with stale endpoint state.
Suggested fix
if (stored_len == (size_t)len &&
strncmp(g_d2dserverlist_d2dserverlist[i].di, di, (size_t)len) == 0) {
- strcpy(g_d2dserverlist_d2dserverlist[i].di, "");
+ free(discovered_server[i]);
+ discovered_server[i] = NULL;
+ memset(&g_d2dserverlist_d2dserverlist[i], 0,
+ sizeof(g_d2dserverlist_d2dserverlist[i]));
return true;
}📝 Committable suggestion
‼️ IMPORTANT
Carefully review the code before committing. Ensure that it accurately replaces the highlighted code, contains no missing lines, and has no issues with indentation. Thoroughly test & benchmark the code to ensure it meets the requirements.
| for (int i = 0; i < MAX_ARRAY; i++) { | |
| OC_PRINTF(" %s %.*s ", g_d2dserverlist_d2dserverlist[i].di, len, di); | |
| if (strncmp(g_d2dserverlist_d2dserverlist[i].di, di, len) == 0) { | |
| size_t stored_len = strlen(g_d2dserverlist_d2dserverlist[i].di); | |
| if (stored_len == (size_t)len && | |
| strncmp(g_d2dserverlist_d2dserverlist[i].di, di, (size_t)len) == 0) { | |
| strcpy(g_d2dserverlist_d2dserverlist[i].di, ""); | |
| return true; | |
| for (int i = 0; i < MAX_ARRAY; i++) { | |
| OC_PRINTF(" %s %.*s ", g_d2dserverlist_d2dserverlist[i].di, len, di); | |
| size_t stored_len = strlen(g_d2dserverlist_d2dserverlist[i].di); | |
| if (stored_len == (size_t)len && | |
| strncmp(g_d2dserverlist_d2dserverlist[i].di, di, (size_t)len) == 0) { | |
| free(discovered_server[i]); | |
| discovered_server[i] = NULL; | |
| memset(&g_d2dserverlist_d2dserverlist[i], 0, | |
| sizeof(g_d2dserverlist_d2dserverlist[i])); | |
| return true; |
🤖 Prompt for AI Agents
Verify each finding against the current code and only fix it if needed.
In `@apps/cloud_proxy.c` around lines 717 - 723, The remove_di() loop currently
only clears g_d2dserverlist_d2dserverlist[i].di, leaking the corresponding
discovered_server[i] endpoint state; update remove_di() so when a DI match is
found you also free or reset the discovered_server entry (e.g., free
discovered_server[i].endpoint or call its cleanup/reset function, zero its
struct, and reset any allocation flags) before returning true, ensuring
discovered_server[i] does not retain stale allocated memory or state for reused
slots.
| oc_resource_t *res = oc_ri_get_app_resources(); | ||
| while (res != NULL) { | ||
| if (strncmp(di, oc_string(res->uri), strlen(di)) == 0) | ||
| if (strncmp(di, oc_string(res->uri), (size_t)di_len) == 0) | ||
| return res; |
There was a problem hiding this comment.
Match registered proxy URIs against /<di>/..., not di at byte 0.
Line 744 compares the query di directly against oc_string(res->uri), but proxied resources are registered as /<uuid>/.... unregister_resources() therefore misses every resource for that device, so DELETE /d2dserverlist?di=... leaves stale proxy resources behind.
Suggested fix
STATIC oc_resource_t *
find_resource(const char *di, int di_len)
{
if (di == NULL || di_len <= 0) {
return NULL;
}
oc_resource_t *res = oc_ri_get_app_resources();
while (res != NULL) {
- if (strncmp(di, oc_string(res->uri), (size_t)di_len) == 0)
+ const char *uri = oc_string(res->uri);
+ if (uri != NULL && uri[0] == '/' &&
+ strncmp(uri + 1, di, (size_t)di_len) == 0 &&
+ (uri[di_len + 1] == '/' || uri[di_len + 1] == '\0')) {
return res;
+ }
res = res->next;
}
- return res;
+ return NULL;
}Also applies to: 766-771
🤖 Prompt for AI Agents
Verify each finding against the current code and only fix it if needed.
In `@apps/cloud_proxy.c` around lines 742 - 745, The loop that searches app
resources (using oc_ri_get_app_resources and in unregister_resources) wrongly
compares di to the resource URI at byte 0; fix it to match registered proxy URIs
of the form "/<di>/...". Specifically, when iterating oc_resource_t *res and
comparing oc_string(res->uri) to di, compare starting at the character after the
leading '/' (e.g. compare oc_string(res->uri) + 1 to di for di_len bytes) and
also require that the character immediately after the matched di in the URI is
'/' (to ensure a "/<di>/" prefix), and apply the same change to the other loop
mentioned (lines 766-771).
|
|
||
| } /* adding current device, e.g. add the resource to the cloud RD */ | ||
| } /* if loop */ | ||
| } /* if loop */ |
There was a problem hiding this comment.
Run clang-format on this hunk.
The formatting job is already failing here; clang-format rewrites the trailing loop comment spacing on this line.
🤖 Prompt for AI Agents
Verify each finding against the current code and only fix it if needed.
In `@apps/cloud_proxy.c` at line 1733, Run clang-format over the reported hunk to
fix spacing of the trailing comment; specifically reformat the line containing
the closing brace and the comment "/* if loop */" so it matches project style
(apply clang-format to the block containing the brace and comment and stage the
updated file apps/cloud_proxy.c). Ensure only formatting changes are committed.
…length in doc comment Agent-Logs-Url: https://github.com/iotivity/iotivity-lite/sessions/b976e73b-b795-4bd6-8741-a7d571f08bf3 Co-authored-by: jkralik <23374974+jkralik@users.noreply.github.com>
