On 12.02.2026 04:23, Marek Marczykowski-Górecki wrote: > Archlinux just updated gcc to 15.2.1+r604+g0b99615a8aef-1 and that > complains about libxl: > > libxl_cpuid.c: In function ‘libxl_cpuid_parse_config_xend’: > libxl_cpuid.c:447:16: error: assignment discards ‘const’ qualifier from > pointer target type [-Werror=discarded-qualifiers] > 447 | endptr = strchr(str, '='); > | ^ > libxl_cpuid.c:452:16: error: assignment discards ‘const’ qualifier from > pointer target type [-Werror=discarded-qualifiers] > 452 | endptr = strchr(str, ','); > | ^ > libxl_cpuid.c:454:20: error: assignment discards ‘const’ qualifier from > pointer target type [-Werror=discarded-qualifiers] > 454 | endptr = strchr(str, 0); > | ^ > libxl_cpuid.c: In function ‘libxl_cpuid_parse_config_xend’: > libxl_cpuid.c:447:16: error: assignment discards ‘const’ qualifier from > pointer target type [-Werror=discarded-qualifiers] > 447 | endptr = strchr(str, '='); > | ^ > libxl_cpuid.c:452:16: error: assignment discards ‘const’ qualifier from > pointer target type [-Werror=discarded-qualifiers] > 452 | endptr = strchr(str, ','); > | ^ > libxl_cpuid.c:454:20: error: assignment discards ‘const’ qualifier from > pointer target type [-Werror=discarded-qualifiers] > 454 | endptr = strchr(str, 0); > | ^ > cc1: all warnings being treated as errors
That's supposed to be happening in C23 mode only, isn't it? Looks like under tools/ we don't set the mode we want to compile in. > Add missing consts. Note in libxl_cpuid_parse_config_xend() non-const > endptr still is needed, to be compatible with the second argument to > strtoul(). Add second variable for this reason. > > Signed-off-by: Marek Marczykowski-Górecki <[email protected]> > --- > When I say "just updated" I really mean it. The update hit me between my > CI saying "ok" and release build few hours later. I guess Xen's CI will > see that only after next refresh of the Arch container (next week?). Hmm, and that would then affect not only stable trees, but also security- only ones. In particular there I'd wonder whether enforcing mode (to be e.g. C99) wouldn't be more appropriate. Anthony? > --- a/tools/libs/light/libxl_cpuid.c > +++ b/tools/libs/light/libxl_cpuid.c > @@ -415,6 +415,7 @@ int libxl_cpuid_parse_config_xend(libxl_cpuid_policy_list > *policy, > const char* str) > { > char *endptr; > + const char *endptrc; Don't know what the (perhaps unwritten) policy in libxl is, but in the hypervisor I'd ask for the variable to be declared in the more narrow scope it's solely used in. > @@ -444,25 +445,25 @@ int > libxl_cpuid_parse_config_xend(libxl_cpuid_policy_list *policy, > return 4; > } > value = str[1] - 'a'; > - endptr = strchr(str, '='); > - if (value > 3 || endptr == NULL) { > + endptrc = strchr(str, '='); > + if (value > 3 || endptrc == NULL) { > return 4; > } > - str = endptr + 1; > - endptr = strchr(str, ','); > - if (endptr == NULL) { > - endptr = strchr(str, 0); > + str = endptrc + 1; > + endptrc = strchr(str, ','); > + if (endptrc == NULL) { > + endptrc = strchr(str, 0); > } > - if (endptr - str != 32) { > + if (endptrc - str != 32) { > return 5; > } > entry->policy[value] = calloc(32 + 1, 1); > strncpy(entry->policy[value], str, 32); > entry->policy[value][32] = 0; > - if (*endptr == 0) { > + if (*endptrc == 0) { > break; > } > - for (str = endptr + 1; *str == ' ' || *str == '\n'; str++); > + for (str = endptrc + 1; *str == ' ' || *str == '\n'; str++); I'd further ask for the semicolon to be moved to its own line. Jan
