|
| 1 | +1.8.2 | 2023-10-26 12:42:44 +0200 |
| 2 | + |
| 3 | + * Bump FreeBSD versions in CI. (Benjamin Bannier, Corelight) |
| 4 | + |
| 5 | + (cherry picked from commit 9afcedcb084bf352a993e86db51261ae63753f17) |
| 6 | + |
| 7 | + * GH-1571: Remove trimming inside individual chunks. (Benjamin Bannier, Corelight) |
| 8 | + |
| 9 | + Trimming `Chunk`s (always from the left) causes a lot of internal work |
| 10 | + with only limited benefit since we manage visibility with `stream::View`s |
| 11 | + on top of `Chunk`s anyway. |
| 12 | + |
| 13 | + This patch removes trimming inside `Chunk`s so now any trimming only |
| 14 | + removes `Chunk`s from `Chain`s, but does not internally change |
| 15 | + individual `Chunk`s anymore. This might lead to slightly increased memory |
| 16 | + use, but callers usually have that data in memory anyway. |
| 17 | + |
| 18 | + Closes #1571. |
| 19 | + |
| 20 | + (cherry picked from commit 625a60745a3e2287f4485d099d382cb20daf8d0a) |
| 21 | + |
| 22 | + * GH-1549: GH-1554: Fix potential infinite loop when trimming data before stream. (Benjamin Bannier, Corelight) |
| 23 | + |
| 24 | + Previously we would trigger an infinite loop if one tried to trim before |
| 25 | + the head chunk of a stream. In praxis this seem to have been no issue |
| 26 | + due to #1549 and us emitting way less calls to trim than possible. |
| 27 | + |
| 28 | + This patch adds an explicit check whether we need to trim anything, and |
| 29 | + exits the low-level function early for such cases. |
| 30 | + |
| 31 | + Closes #1554. |
| 32 | + |
| 33 | + (cherry picked from commit bc888f5bf82c18cc2a1e73f04294985d54d257c3) |
| 34 | + |
| 35 | + * GH-1550: Replace recursive deletion with explicit loop to avoid stack overflow. (Benjamin Bannier, Corelight) |
| 36 | + |
| 37 | + Closes #1550. |
| 38 | + |
| 39 | + (cherry picked from commit e190a7875215801af37bbb614ca7cd9b43aad84b) |
| 40 | + |
| 41 | + * GH-1549: Add feature guards to accesses of a unit's `__position`. (Benjamin Bannier, Corelight) |
| 42 | + |
| 43 | + Access of `__position` triggers a random access functionality. In order |
| 44 | + to distinguish our internal uses from accesses due to user code, most |
| 45 | + access in our generated code should be guarded with a feature constant |
| 46 | + (`if` or ternary). |
| 47 | + |
| 48 | + In this patch add proper guards for a couple instances where we did not |
| 49 | + do that correctly. That mishap caused all units with containers to be |
| 50 | + random access (even the root unit) which in turn could have lead to |
| 51 | + e.g., unbounded memory growth, or runtime overhead due to generation and |
| 52 | + execution of unneeded code, or expensive cleanup on very large untrimmed |
| 53 | + inputs. |
| 54 | + |
| 55 | + Closes #1549. |
| 56 | + |
| 57 | + (cherry picked from commit acd3844f0c94e2c6025f89453e6a0519f9f2025e) |
| 58 | + |
| 59 | + * Artificially limit the number of open files. (Benjamin Bannier, Corelight) |
| 60 | + |
| 61 | + This works around a silent failure in reproc where it would refuse to |
| 62 | + run on systems which huge rlimits for the number of open files. We have |
| 63 | + seen this hit on huge production boxes. |
| 64 | + |
| 65 | + (cherry picked from commit 273e44281f0d890c183e7fd7ee128db44c16db67) |
| 66 | + |
| 67 | + * GH-1478: Add regression test for #1478. (Benjamin Bannier, Corelight) |
| 68 | + |
| 69 | + Closes #1478. |
| 70 | + |
| 71 | + (cherry picked from commit f4ef4c572c21b4d97499a20f7a931b739ecb46d2) |
| 72 | + |
| 73 | + * Add begin to parser state. (Benjamin Bannier, Corelight) |
| 74 | + |
| 75 | + This patch adds the current begin position to the parser state, and |
| 76 | + makes the corresponding changes to generated parser functions so it is |
| 77 | + passed down. |
| 78 | + |
| 79 | + We already modelled the semantic beginning of the input in the unit, but |
| 80 | + had no reliable way to keep this up-to-date across non-unit contexts |
| 81 | + like `&parse-from`. This would then for certain setups lead to generated |
| 82 | + code where `input` and `position` would point to different inputs which in |
| 83 | + turn caused `offset` (modelled as `position - input`) to be incorrect. |
| 84 | + |
| 85 | + (cherry picked from commit 35aa4266c8bf37ebff3df5ec17241171e42e7cec) |
| 86 | + |
| 87 | + * Expand validator error message. (Benjamin Bannier, Corelight) |
| 88 | + |
| 89 | + (cherry picked from commit d77775c81e4128603b899ba9bd3817a90a8bd58d) |
| 90 | + |
| 91 | + * Disable a few newer clang-tidy categories. (Benjamin Bannier, Corelight) |
| 92 | + |
| 93 | + The options disabled here and triggered in newer versions of clang-tidy. |
| 94 | + |
| 95 | + (cherry picked from commit 08998a7192abaeb4ef1967a28eadcba622be0bb5) |
| 96 | + |
| 97 | + * Drop `-noall_load` linker option. (Benjamin Bannier, Corelight) |
| 98 | + |
| 99 | + We added this linker option on macos. This option was already obsolete, |
| 100 | + e.g., in the `ld` manpage: |
| 101 | + |
| 102 | + ``` |
| 103 | + -noall_load |
| 104 | + This is the default. This option is obsolete. |
| 105 | + ``` |
| 106 | + |
| 107 | + Newer versions of xcode do not know this option anymore and instead |
| 108 | + generate a hard error. |
| 109 | + |
| 110 | + (cherry picked from commit 265581faeaf4bebfb2255ee4a39a87ffba206466) |
| 111 | + |
| 112 | + * Declare Spicy pygments extension as parallel-safe. [skip CI] (Benjamin Bannier, Corelight) |
| 113 | + |
| 114 | + We previously would not declare that the Spicy pygments highlighter is |
| 115 | + safe to execute in parallel (reading or writing of sources). Sphinx then |
| 116 | + assumed that the extension was not safe to run in parallel and instead |
| 117 | + ran jobs sequentially. |
| 118 | + |
| 119 | + This patch declares the extension as able to execute in parallel. Since |
| 120 | + the extension does not manage any external state this is safe. |
| 121 | + |
| 122 | + (cherry picked from commit 22e9f729f423d0e43b788d48d824377ceaaf8ede) |
| 123 | + |
| 124 | + * Use find_package(Python) with version. (Arne Welzel, Corelight) |
| 125 | + |
| 126 | + Zeek's configure sets Python_EXECUTABLE has hint, but Spicy is using |
| 127 | + find_package(Python3) and would only use Python3_EXECUTABLE as hint. |
| 128 | + This results in Spicy finding a different (the default) Python executable |
| 129 | + when configuring Zeek with --with-python=/opt/custom/bin/python3. |
| 130 | + |
| 131 | + Switch Spicy over to use find_package(Python) and add the minimum |
| 132 | + version so it knows to look for Python3. |
| 133 | + |
| 134 | + (cherry picked from commit 32d8bbd85d03fe821d77729d3976dd0e37180474) |
| 135 | + |
1 | 136 | 1.8.1 | 2023-07-04 17:16:32 +0200 |
2 | 137 |
|
3 | 138 | * GH-1475: Revert #1439. (Benjamin Bannier, Corelight) |
|
0 commit comments