ticket #TSP-8629412 bug fix #3187

dkjsone · 2025-10-17T06:53:52Z

A total of 8 vulnerabilities were fixed, including 5 integer overflows, 2 division-by-zero crashes, and 1 null pointer crash.

v4.9.3-rc2 wellspring changes

Update upload-artifact/download-artifact for github actions

Fix out-of-tree builds generating netcdf_json.h and netcdf_proplist.h

Update error code list.

…ss or fail.

Capture ac-based log artifacts

Clean up a couple small things while I'm looking at them.

…ata/netcdf-c/3067

## Primary Changes 1. The ncdap_test/test_manyurls.c build in ncdap_test/Makefile.am caused some makefile warnings. These were fixed. 2. Make the test program ncdap_test/test_manyurls.c be operable. 3. After 1 and 2, it turns out that the test case is probably useless, so disable it. ## Secondary Changes 1. Fix the fact that the --enable-dap-long-tests option was not generating a corresponding NETCDF_ENABLE_DAP_LONG_TESTS flag in config.h. 2. Add NETCDF_ENABLE_DAP_LONG_TESTS to config.h.cmake.in to make it visible. 3. Discovered and fixed an obscure error in libdispatch/nclist.c function nameed nclistsetalloc. 4. Add support for CURLOPT_VERBOSE (as an environment variable) to libsrc/httpio.c.

re: Issue Unidata#3184 When ncdump attempts to find a dimension's Fully Qualified Name (FQN), it uses a two step process. 1. Search up the group tree starting at the group in which the dimension the referenced. 2. If not found, then do a breadth first search of the whole group tree looking for the dimension. Once located, a FQN is constructed with respect to the group in which the dimension was found. It turns out there was a bug in the ncdump function "searchgroupdim" such that it always returned the top level group as the one containing the dimension. This, of course is erroneous for nested dimension. So the "searchgroupdim" function was fixed to properly fail if the dimension was not actually in the root group.

Fix bug in ncdump when printing dimension FQN

Fix the problems around ncdap_test/test_manyurls.c.

CLAassistant · 2025-10-17T06:53:59Z

Thank you for your submission! We really appreciate it. Like many open source projects, we ask that you all sign our Contributor License Agreement before we can accept your contribution.
4 out of 5 committers have signed the CLA.

✅ WardF
✅ Dave-Allured
✅ dkjsone
✅ DennisHeimbigner
❌ johnwparent
_{You have signed the CLA already but the status is still pending? Let us recheck it.}

DennisHeimbigner

This line and similar lines use size_t as a constant, but it fails in many cases.
What is the supposed value of size_t as a constant?

#define SIZE_MAX ((size_t)-1)

DennisHeimbigner · 2025-10-17T20:00:48Z

Can you please annotate the code changes to indicate what error they are fixing?

dkjsone · 2025-10-18T06:32:36Z

This line and similar lines use size_t as a constant, but it fails in many cases. What is the supposed value of size_t as a constant?
#define SIZE_MAX ((size_t)-1)

(size_t)-1 is used to obtain the platform-independent maximum value of size_t. On 32-bit systems, it is 0xFFFFFFFF, and on 64-bit systems, it is 0xFFFFFFFFFFFFFFFF.
Initially, I did not define this macro, and during Checks, the macOS platform encountered an issue of the macro being undefined. Therefore, I added this macro definition.

dkjsone · 2025-10-18T06:37:41Z

Can you please annotate the code changes to indicate what error they are fixing?

The changes I made to the code are relatively minor and quite similar, pertaining to routine security checks. I believe comments may not be strictly necessary in this case.
I have provided a detailed description of the issue and the reproduction method in this link. If there are any questions regarding the vulnerability, you can refer to this link for clarification.
https://support.unidata.ucar.edu/tickets/view/TSP-8629412

WardF · 2025-10-23T17:13:13Z

Hi @dkjsone, thanks for submitting this. I'm reviewing it now as well. In the meantime, do you mind accepting the CLA here so that we can clear that 'pending check'? Thanks!

dkjsone · 2025-10-24T07:46:29Z

Sorry, my mistake — I'm going to close this one and start over with a new pull request.

dkjsone · 2025-10-24T08:51:08Z

Please review this new pull request.
#3191

WardF and others added 30 commits December 18, 2024 16:17

A more reasonable, simpler approach

23ed79a

Allow error catching.

f9ce9e3

Revert to hdf5 1.14.3

abfb2c4

Add more diagnostic info

4b79492

Additional cache checking.

2db3eb3

Updated to remove custom mpich installation.

28478b3

Following a hunch

636dc4a

Update cache

4a82eb9

Correct misuse of github runner variable.

4a26307

Merge pull request Unidata#3052 from Unidata/v4.9.3-rc2-wellspring.wif

7a1417d

v4.9.3-rc2 wellspring changes

Merging in main branch.

e9ef44e

Add job name to avoid namespace collision

6632468

Correct logic to avoid a namespace conflict.

b48480f

Merge pull request Unidata#3065 from WardF/update-artifact-action.wif

b84d4eb

Update upload-artifact/download-artifact for github actions

Merge branch 'main' of https://github.com/Unidata/netcdf-c into srcdir

6145523

Merge pull request Unidata#3060 from opoplawski/srcdir

eac7d2a

Fix out-of-tree builds generating netcdf_json.h and netcdf_proplist.h

Update all-error-codes.md in support of Unidata#3072

b282699

Remove final error code placeholder.

137ea36

Merge pull request Unidata#3073 from WardF/gh3072-fix-error-codes.wif

4990485

Update error code list.

Add stanza so that autoconf-based log files artifacts are captured pa…

a1441d2

…ss or fail.

Correct indentation error.

bb85ee3

Merge pull request Unidata#3074 from WardF/ac-artifacts.wif

7afe659

Capture ac-based log artifacts

Clean up a couple small things while I'm looking at them.

691b906

Merge branch 'main' into tiny-cleanup.wif

fe0336e

Avoid name conflict in artifact upload.

29d9021

Merge pull request Unidata#3076 from WardF/tiny-cleanup.wif

d4b7ff6

Clean up a couple small things while I'm looking at them.

Update default chunk_cache_size in support of https://github.com/Unid…

3042579

…ata/netcdf-c/3067

Add cache fix to configure.ac

e314d0b

Fix an error in the cygwin workflows

4ae8dfe

Attempted fix for a cygwin artifact name collision

3ade8e8

DennisHeimbigner and others added 15 commits October 11, 2025 18:47

nclist

e9f4c32

alloc fix

e0ef3f3

test10

3098520

push1

64848f1

test1

bd0743b

test2

edee9bc

ckp

371de07

test1

7dacee1

unpush

97d6fc6

Merge pull request Unidata#3185 from DennisHeimbigner/dimidsearch.dmh

51a55bc

Fix bug in ncdump when printing dimension FQN

Merge branch 'main' into manyurls.dmh

5171dd6

Merge pull request Unidata#3182 from DennisHeimbigner/manyurls.dmh

b5c9ba0

Fix the problems around ncdap_test/test_manyurls.c.

Merge branch 'main' into bugfix

4a42dae

dkjsone requested review from DennisHeimbigner and WardF as code owners October 17, 2025 06:53

dkjsone added 2 commits October 17, 2025 15:30

fix macro MAX_SIZE

95e7b52

Fix macro SIZE_MAX

48bad22

DennisHeimbigner reviewed Oct 17, 2025

View reviewed changes

dkjsone force-pushed the bugfix branch from 2ae43c5 to 48bad22 Compare October 24, 2025 07:41

dkjsone closed this Oct 24, 2025

dkjsone deleted the bugfix branch October 24, 2025 08:45

Provide feedback

Saved searches

Use saved searches to filter your results more quickly

Uh oh!

ticket #TSP-8629412 bug fix #3187

ticket #TSP-8629412 bug fix #3187

Uh oh!

dkjsone commented Oct 17, 2025

Uh oh!

CLAassistant commented Oct 17, 2025 •

edited

Loading

Uh oh!

DennisHeimbigner left a comment

Uh oh!

DennisHeimbigner commented Oct 17, 2025

Uh oh!

dkjsone commented Oct 18, 2025

Uh oh!

dkjsone commented Oct 18, 2025

Uh oh!

WardF commented Oct 23, 2025

Uh oh!

dkjsone commented Oct 24, 2025

Uh oh!

dkjsone commented Oct 24, 2025

Uh oh!

Reviewers

Assignees

Labels

Projects

Milestone

Development

Uh oh!

13 participants

ticket #TSP-8629412 bug fix #3187

ticket #TSP-8629412 bug fix #3187

Uh oh!

Conversation

dkjsone commented Oct 17, 2025

Uh oh!

CLAassistant commented Oct 17, 2025 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Uh oh!

DennisHeimbigner left a comment

Choose a reason for hiding this comment

Uh oh!

DennisHeimbigner commented Oct 17, 2025

Uh oh!

dkjsone commented Oct 18, 2025

Uh oh!

dkjsone commented Oct 18, 2025

Uh oh!

WardF commented Oct 23, 2025

Uh oh!

dkjsone commented Oct 24, 2025

Uh oh!

dkjsone commented Oct 24, 2025

Uh oh!

Reviewers

Assignees

Labels

Projects

Milestone

Development

Uh oh!

13 participants

CLAassistant commented Oct 17, 2025 •

edited

Loading