[PAL] Fix memory PAL regression test for UBSan

[dimakuv]

Description of the changes

Newer Clang versions (v18 and newer) added more UBSan checks that triggered UBSan on PAL regression tests. In particular, UBSan added the check "Indirect call of a function through a function pointer of the wrong type". UBSan expects all functions which can be indirectly called to be instrumented with two magic metadata values, located right-before the function in address space.

The memory test however does not add this required metadata to a dummy generated-on-the-fly function (it simply allocates some pages and copies the asm code at the beginning of the first page). Instead of adding UBSan-required metadata, we simply disable this particular check (it's a test after all, not core Gramine functionality).

References:

• https://maskray.me/blog/2022-12-18-control-flow-integrity#fsanitizefunction
• https://llvm.org/docs/LangRef.html#func-sanitize-metadata

Reported by @woju.

How to test this PR?

Run PAL regression tests on Ubuntu 24.04 + UBSan enabled.

---

This change is 

[kailun-qin]

Reviewed 1 of 1 files at r1, all commit messages.
Reviewable status: all files reviewed, 2 unresolved discussions, not enough approvals from maintainers (2 more required), not enough approvals from different teams (1 more required, approved so far: Intel) (waiting on @dimakuv)

---

-- commits line 4 at r1:
Isn't it v17? (according to https://maskray.me/blog/2022-12-18-control-flow-integrity#fsanitizefunction)

Code quote:

v18

---

-- commits line 4 at r1:
Isn't it (also) because since v17-fsanitize=function was allowed on C code, while previously it was C++ only (based on https://reviews.llvm.org/D148827)?

Code quote:

added more UBSan checks

[dimakuv]

Reviewable status: all files reviewed, 2 unresolved discussions, not enough approvals from maintainers (2 more required), not enough approvals from different teams (1 more required, approved so far: Intel) (waiting on @kailun-qin)

---

-- commits line 4 at r1:

Previously, kailun-qin (Kailun Qin) wrote…

Isn't it v17? (according to https://maskray.me/blog/2022-12-18-control-flow-integrity#fsanitizefunction)

Done.

---

-- commits line 4 at r1:

Previously, kailun-qin (Kailun Qin) wrote…

Isn't it (also) because since v17-fsanitize=function was allowed on C code, while previously it was C++ only (based on https://reviews.llvm.org/D148827)?

Done. Rephrased.

[kailun-qin]

Reviewed all commit messages.
Reviewable status: all files reviewed, all discussions resolved, not enough approvals from maintainers (1 more required), not enough approvals from different teams (1 more required, approved so far: Intel)

[mkow]

Reviewed 1 of 1 files at r1, all commit messages.
Reviewable status: all files reviewed, 1 unresolved discussion, not enough approvals from maintainers (1 more required), not enough approvals from different teams (1 more required, approved so far: Intel) (waiting on @dimakuv)

---

pal/regression/memory.c line 96 at r2 (raw file):

 * to be instrumented with two magic metadata values, located right-before the function in address
 * space. The below code does *not* add metadata (it simply allocates some pages and copies the asm
 * code at the beginning of the first page). Instead of adding UBSan-required metadata, we simply

"asm code" -> "machine code"

[dimakuv]

Reviewable status: 0 of 1 files reviewed, 1 unresolved discussion, not enough approvals from maintainers (2 more required), not enough approvals from different teams (1 more required, approved so far: Intel) (waiting on @kailun-qin and @mkow)

---

pal/regression/memory.c line 96 at r2 (raw file):

Previously, mkow (Michał Kowalczyk) wrote…

"asm code" -> "machine code"

Done. Also in the commit message.

[mkow]

Reviewed 1 of 1 files at r3, all commit messages.
Reviewable status: all files reviewed, all discussions resolved, not enough approvals from maintainers (1 more required)

[kailun-qin]

Reviewed 1 of 1 files at r3, all commit messages.
Reviewable status: complete! all files reviewed, all discussions resolved

[mkow]

Reviewed 1 of 1 files at r4, all commit messages.
Reviewable status: complete! all files reviewed, all discussions resolved