Skip to contentSkip to conversation

bpf, riscv: Make arena support depend on ZACAS

PATCH Superseded Chen Pei 2 participants 2 messages Sep 1

Subsystems (4)

v1 of 2 Superseded Latest is v2 →

lore
Share & tools
lore

Chen Pei

author
#0 lore raw

PATCH 1/1 bpf, riscv: Make arena support depend on ZACAS

The arena range tree allocates its nodes with kmalloc_nolock() since commit f8c67d8550ee ("bpf: Use kmalloc_nolock() in range tree"). kmalloc_nolock() requires slab caches with cmpxchg128 support (__CMPXCHG_DOUBLE); on riscv cmpxchg128 is provided by the ZACAS extension. On systems without ZACAS every arena map creation fails with a misleading -ENOMEM.

Report the missing support instead: make bpf_jit_supports_arena() return system_has_cmpxchg128(), so arena map creation fails with -EOPNOTSUPP on systems without ZACAS. This matches how arena BPF_CMPXCHG instructions are already gated on ZACAS in bpf_jit_supports_insn().

Fixes f8c67d8550ee ("bpf: Use kmalloc_nolock() in range tree")

---
This issue was reported by sashiko-bot:
https://sashiko.dev/#/patchset/[email protected]?part=1

Question for reviewers: should the arena selftests gate on ZACAS, e.g. probing it via riscv_hwprobe() (RISCV_ISA_EXT_ZACAS) and SKIPping cleanly on systems without the extension?

 arch/riscv/net/bpf_jit_comp64.c | 6 +++++-
 1 file changed, 5 insertions(+), 1 deletion(-)
arch/riscv/net/bpf_jit_comp64.c +5 -1
index 74efe4b138d2..0061c1d4ed98 100644
@@ -2128,7 +2128,11 @@ bool bpf_jit_supports_ptr_xchg(void)
 
 bool bpf_jit_supports_arena(void)
 {
-	return true;
+	/*
+	 * The arena range tree uses kmalloc_nolock(), which needs
+	 * cmpxchg128, provided by ZACAS on riscv.
+	 */
+	return system_has_cmpxchg128();
 }
 
 bool bpf_jit_supports_insn(struct bpf_insn *insn, bool in_arena)
Signature · 1 line
2.50.1
#1 lore raw
Thank you for your contribution! Sashiko AI review found 1 potential issue(s) to consider: - [Low] The unconditional use of `system_has_cmpxchg128()` in `bpf_jit_supports_arena()` causes a build…

Thank you for your contribution! Sashiko AI review found 1 potential issue(s) to consider:
- [Low] The unconditional use of system_has_cmpxchg128() in bpf_jit_supports_arena() causes a build failure when CONFIG_RISCV_ISA_ZACAS or CONFIG_TOOLCHAIN_HAS_ZACAS is disabled. --

commit c73d679b84188e98d5cff0171b751fa3f4ce5627
Author: Chen Pei <[email protected]>

bpf, riscv: Make arena support depend on ZACAS

This patch updates bpf_jit_supports_arena() to return the result of system_has_cmpxchg128(). This ensures that BPF arena map creation fails with -EOPNOTSUPP on RISC-V systems that do not have the ZACAS extension.

Quoted 5 lines
arch/riscv/net/bpf_jit_comp64.c +0 -0
index e7378be171a9f..a5870d386f983 100644

arch/riscv/net/bpf_jit_comp64.c bool bpf_jit_supports_ptr_xchg(void) · line 2148

Quoted 9 lines
arch/riscv/net/bpf_jit_comp64.c +5 -1
@@ -2141,7 +2141,11 @@ bool bpf_jit_supports_ptr_xchg(void)
 
 bool bpf_jit_supports_arena(void)
 {
-	return true;
+	/*
+	 * The arena range tree uses kmalloc_nolock(), which needs
+	 * cmpxchg128, provided by ZACAS on riscv.
+	 */
+	return system_has_cmpxchg128();

[Severity: Low]
Does this code cause a build regression when compiling without ZACAS support?

Looking at arch/riscv/include/asm/cmpxchg.h, system_has_cmpxchg128() is conditionally defined without a fallback:

#if defined(CONFIG_64BIT) && defined(CONFIG_RISCV_ISA_ZACAS) && defined(CONFIG_TOOLCHAIN_HAS_ZACAS)
#define system_has_cmpxchg128() riscv_has_extension_unlikely(...)
...
#endif

If a kernel is built with CONFIG_BPF_JIT=y but CONFIG_RISCV_ISA_ZACAS=n (or CONFIG_TOOLCHAIN_HAS_ZACAS=n), will the unconditional use of system_has_cmpxchg128() in bpf_jit_supports_arena() result in an implicit declaration error?

Quoted 1 line
arch/riscv/net/bpf_jit_comp64.c +0 -0
 }
Signature · 1 line
Sashiko AI review · https://sashiko.dev/#/patchset/[email protected]?part=1