| Patch | Description | Author | Forwarded | Bugs | Origin | Last update |
|---|---|---|---|---|---|---|
| Fix-writing-to-the-database-with-ProtectSystem-strict.patch | Fix writing to the database with ProtectSystem=strict Fixes https://github.com/hughsie/colord/issues/166 (cherry picked from commit 08a32b2379fb5582f4312e59bf51a2823df56276) |
Richard Hughes <richard@hughsie.com> | no | upstream after 1.4.7 | 2024-01-29 | |
| Fix-USB-scanners-not-working-with-RestrictAddressFamilies.patch | Fix USB scanners not working with RestrictAddressFamilies colord-sane scanner drivers using libusb can't initialize properly with RestrictAddressFamilies set to AF_UNIX. Remove that line to ensure those can work properly. This also avoids a crash in HPLIP due to unchecked calls to libusb_init(). Fixes #165 (cherry picked from commit 9283abd9c00468edb94d2a06d6fa3681cae2700d) |
Ferdinand Bachmann <ferdinand.bachmann@yrlf.at> | no | upstream after 1.4.7 | 2024-01-30 | |
| fix-NULL-passed-to-free-with-sqlite3-error_msg-pointers.patch | fix NULL passed to free with sqlite3 error_msg pointers when an error does not happen, and an error_msg pointer is passed, sqlite does not touch it. that means this pointer is uninitialised (=NULL) which violates the constraints of g_autofree ("the variable must be initialized"). this is then passed to g_free and crashes in tests. use a regular pointer and free it manually on error, after sqlite writes to it after setting it with sqlite_malloc. fixes https://github.com/hughsie/colord/issues/163 (cherry picked from commit 7435d9c23a1d3bc138dcae0e05ea06fb244568fe) |
psykose <alice@ayaya.dev> | no | upstream after 1.4.7 | 2024-02-08 | |
| use-char-instead-of-gchar-for-pointers-passed-to-sqlite3_.patch | use char * instead of gchar * for pointers passed to sqlite3_exec (cherry picked from commit d8076a16ea1f4437cca3799d414faf40cddae955) |
psykose <alice@ayaya.dev> | no | upstream after 1.4.7 | 2024-02-08 | |
| use-g_autoptr-again-but-with-a-custom-CLEANUP_FUNC.patch | use g_autoptr again but with a custom CLEANUP_FUNC my prior analysis was completely wrong, and the original crash was seen because sqlite3_malloc pointers passed to g_free cause a mismatched malloc/free, not because of lack of initialisation. use autoptr with an overriden CLEANUP_FUNC instead. for cd_mapping_db_load/cd_device_db_remove, keep using manual sqlite3_free since we reuse the same error_msg multiple times. (cherry picked from commit f4fdb66d0399e3223c068c663e7621ea19b71a7b) |
psykose <alice@ayaya.dev> | no | upstream after 1.4.7 | 2024-02-12 |