Recent GCC (seen on 9.3.1) complain when compiling stap modules for 32-bit systems. The behaviour that's now being flagged is well-known, unobjectionable and fairly difficult to get rid of -- it's based on the fact that stap scalar values are always 64-bit words regardless of the type being store in them. Most likely, we should suppress the warnings in question, but I'm filing a PR to keep track of the decision in case there are any surprises. Example: as_root /notnfs/smakarov/stap-checkout/stap_install/bin/stap --rlimit-as=1 --rlimit-as=99999999999\ 9 -wp4 /notnfs/smakarov/stap-checkout/testsuite/systemtap.base/rlimit.stp OUT /tmp/stapwwHXqw/stap_2fcc2a5ab281c9226223b81a92db655d_2051647_src.c: In function 'function___g\ lobal___get_kregs__overload_0': /tmp/stapwwHXqw/stap_2fcc2a5ab281c9226223b81a92db655d_2051647_src.c:328917:33: error: cast from po\ inter to integer of different size [-Werror=pointer-to-int-cast] 328917 | /* guru */ STAP_RETVALUE = (int64_t)CONTEXT->kregs; | ^ /tmp/stapwwHXqw/stap_2fcc2a5ab281c9226223b81a92db655d_2051647_src.c: In function 'function___globa\ l___set_kregs__overload_0': /tmp/stapwwHXqw/stap_2fcc2a5ab281c9226223b81a92db655d_2051647_src.c:328956:34: error: cast to poin\ ter from integer of different size [-Werror=int-to-pointer-cast] 328956 | /* guru */ CONTEXT->kregs = (struct pt_regs*)STAP_ARG_regs; | ^ /tmp/stapwwHXqw/stap_2fcc2a5ab281c9226223b81a92db655d_2051647_src.c: In function 'function___priva\ te__notnfs_smakarov_stap_checkout_testsuite_systemtap_base_rlimit_stp__dwarf_tvar_get_new_rev_269': /tmp/stapwwHXqw/stap_2fcc2a5ab281c9226223b81a92db655d_2051647_src.c:369039:111: error: conversion \ from 'long long unsigned int' to 'long unsigned int' changes value from '18446744070503413664' to \ '1088829344' [-Werror=overflow] 369039 | l->__retvalue = (((((int64_t) (({ unsigned long addr = 0; addr = _stp_kmodule_relocate \ ("kernel", "_stext", 0xffffffff40e637a0); addr; }))))) & (((int64_t)4294967295LL))); | \ ^~~~~~~~~~~~~~~~~~ /tmp/stapwwHXqw/stap_2fcc2a5ab281c9226223b81a92db655d_2051647_src.c: In function 'function___priva\ te__notnfs_smakarov_stap_checkout_testsuite_systemtap_base_rlimit_stp__dwarf_tvar_get_patch_268': /tmp/stapwwHXqw/stap_2fcc2a5ab281c9226223b81a92db655d_2051647_src.c:376520:111: error: conversion \ from 'long long unsigned int' to 'long unsigned int' changes value from '18446744070503409568' to \ '1088825248' [-Werror=overflow] 376520 | l->__retvalue = (((((int64_t) (({ unsigned long addr = 0; addr = _stp_kmodule_relocate \ ("kernel", "_stext", 0xffffffff40e627a0); addr; }))))) & (((int64_t)4294967295LL))); | \ ^~~~~~~~~~~~~~~~~~
The primary source of errors -Wpointer<->int-cast has been disabled in commit 4f9715d3b8 Remaining issues I can spot in the test output: - -Werror=overflow in the rlimit.exp testcase (see previous post) - -Wpointer-to-int-cast warnings when compiling java/HelperSDT.c CC HelperSDT.o /notnfs/smakarov/stap-checkout/java/HelperSDT.c: In function ‘determine_java_type’: /notnfs/smakarov/stap-checkout/java/HelperSDT.c:50:12: warning: cast from pointer to integer of di\ fferent size [-Wpointer-to-int-cast] 50 | return (int64_t) strdup("(null)"); /* need a real string to avoid user_string_warn get\ ting upset */ | ^ /notnfs/smakarov/stap-checkout/java/HelperSDT.c:90:10: warning: cast from pointer to integer of di\ fferent size [-Wpointer-to-int-cast] 90 | return (int64_t) get_java_tostring(env, _arg); | ^ /notnfs/smakarov/stap-checkout/java/HelperSDT.c: In function ‘free_sargs’: /notnfs/smakarov/stap-checkout/java/HelperSDT.c:109:12: warning: cast to pointer from integer of d\ ifferent size [-Wint-to-pointer-cast] 109 | free((void *) sargs[i]); | ^ CCLD libHelperSDT.so