[DTrace-devel] [PATCH] cg: rework data size to BPF size conversion and improve error reporting

[Eugene Loh]

Reviewed-by: Eugene Loh <eugene.loh at oracle.com>

I don't know in what order patches will appear, but there might still be 
a vestigial assert() in the ip provider to be removed.

Also, in:

+/*
+ * Variant of store and load macros that takes an actual data size and call
+ * ldstsz(sz) to determine the instruction size specifier.
+ */
+#define BPF_STOREX(sz, dst, ofs, src)                                  \
+               BPF_STORE(bpf_ldst_size(sz, 1), (dst), (ofs), (src))
+#define BPF_STOREX_IMM(sz, dst, ofs, src)                              \
+               BPF_STORE_IMM(bpf_ldst_size(sz, 1), (dst), (ofs), (src))
+#define BPF_LOADX(sz, dst, src, ofs)                                   \
+               BPF_LOAD(bpf_ldst_size(sz, 0), (dst), (src), (ofs))
+

maybe that comment about ldstsz(sz) should be bpf_ldst_size(). Also, 
grammatically, the comment should be consistent about singular/plural.  
Either "variant takes calls" or "take call". Maybe s/Variant/Variants/ 
and s/takes/take/.