> > > > > > > > > > Hi Qi, > > > > > > > > > > Thanks for these patches, and also for rebasing to the latest! > > > > > > > > > > ciniparser.c adds a new warning (see below). Could you fix that > > > > > up in a new patch on top of the initial import (i.e. we retain > > > > > the as-is import, and make a record of what we changed). > > > > > > > > > > Looks like the -Wformat-truncation= warnings were first > > > > > introduced in gcc-7.1, but only became really zealous after > > > > > gcc-8.1. The right solution here might just be to suppress it by > > > > > adding a -Wno-format- truncation to CFLAGS, but I'm open to > > > > > considering other alternatives. > > > > > > > > That warning looks pretty handy. > > > > > > > > > > > > > > $ gcc --version > > > > > gcc (GCC) 8.2.1 20181105 (Red Hat 8.2.1-5) > > > > > > > > > > ccan/ciniparser/ciniparser.c: In function ‘ciniparser_load’: > > > > > ccan/ciniparser/ciniparser.c:442:39: warning: ‘%s’ directive > > > > > output may be truncated writing up to 1024 bytes into a region > > > > > of size between 0 and 1024 [-Wformat-truncation=] > > > > > snprintf(tmp, ASCIILINESZ + 1, "%s:%s", section, key); > > > > > ^~ ~~~ > > > > > In file included from /usr/include/stdio.h:873, > > > > > from ./ccan/ciniparser/ciniparser.h:39, > > > > > from ccan/ciniparser/ciniparser.c:36: > > > > > /usr/include/bits/stdio2.h:67:10: note: > > > > > ‘__builtin___snprintf_chk’ output between 2 and 2050 bytes into > > > > > a destination of size 1025 > > > > > return __builtin___snprintf_chk (__s, __n, > > > > > __USE_FORTIFY_LEVEL - 1, > > > > > > > > > > ^~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ > > > > > ~~~~~~ > > > > > __bos (__s), __fmt, __va_arg_pack ()); > > > > > ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ > > > > > > > > Since it's an snprintf without the error return being checked, > > > > perhaps it would be happier as an sprintf with a precision > > > > specified to limit the output? > > > > > > That would work, but 'section' and 'key' are both ASCIILINESZ+1, so > > > how would we specify precision for both the %s specifiers? i.e. > > > would we have to make up an artificial split? > > > > Good point. No way to do it without variable precision. However, > > looking closer I think the warning only triggers when the return value > > is not checked, so perhaps better to just add error handling to that > > case. > > Yep that seems reasonable to me.
Hi, Thank you for your comments. I updated the gcc version in my local environment and got the warning messages. I will make a new version to fix them. Thank you. Qi _______________________________________________ Linux-nvdimm mailing list Linux-nvdimm@lists.01.org https://lists.01.org/mailman/listinfo/linux-nvdimm
