Re: amiga-fdisk: ftbfs with GCC-14
Hi Christian,
On Thu, 2024-08-15 at 22:56 +0200, Christian T. Steigies wrote:
> gcc -g -O2 -Werror=implicit-function-declaration -ffile-prefix-map=/home/cts/salsa/amiga-fdisk=. -fstack-protector-strong -fstack-clash-protection -Wformat -Werror=format-security -fcf-protection -Wall -pedantic -DUSE_READLINE=yes -I./include -Wdate-time -D_FORTIFY_SOURCE=2 -c -o amigastuff.o amigastuff.c
> amigastuff.c: In function ‘rigiddisk_reorg’:
> amigastuff.c:859:20: warning: pointer targets in assignment from ‘LONG *’ {aka ‘long int *’} to ‘ULONG *’ {aka ‘long unsigned int *’} differ in signedness [-Wpointer-sign]
> 859 | crk=&(FSHB(curr)->fhb_SegListBlocks);
> | ^
In this case, you should just declare your variable such that it matches the function call.
Switching between signed and unsigned without proper conversion is not a good idea in general.
> gcc -g -O2 -Werror=implicit-function-declaration -ffile-prefix-map=/home/cts/salsa/amiga-fdisk=. -fstack-protector-strong -fstack-clash-protection -Wformat -Werror=format-security -fcf-protection -Wall -pedantic -DUSE_READLINE=yes -I./include -Wdate-time -D_FORTIFY_SOURCE=2 -c -o fdisk.o fdisk.c
> fdisk.c: In function ‘atonum’:
> fdisk.c:85:33: warning: format ‘%x’ expects argument of type ‘unsigned int *’, but argument 3 has type ‘int *’ [-Wformat=]
> 85 | sscanf(s + 2, "%x", &n);
> | ~^ ~~
> | | |
> | | int *
> | unsigned int *
> | %x
> fdisk.c:87:33: warning: format ‘%o’ expects argument of type ‘unsigned int *’, but argument 3 has type ‘int *’ [-Wformat=]
> 87 | sscanf(s + 1, "%o", &n);
> | ~^ ~~
> | | |
> | | int *
> | unsigned int *
> | %o
>
> I tried defining n in fdisk.c as unsigned int but this caused even more
> warnings in other places. Do I have to cast the arguments in sscanf to
> unsigned int? It wants "unsigned int *", is it as simple as this or is
> there a better solution?
In this case, I would just recommend to use the proper format specifiers, see:
> https://en.cppreference.com/w/c/io/fprintf
> https://www.ibm.com/docs/en/zos/2.4.0?topic=files-inttypesh-io-format-integer-types
I fixed similar bugs in powerpc-utils in the past, see:
> https://github.com/ibm-power-utilities/powerpc-utils/commit/b46743ca68b6a06a2c82de4048d94d3d5191717f
It can be a bit tricky to find the proper format specifier. When testing the change,
make sure to test build on both 32- and 64-bit systems as well as little- and big-
endian.
I would recommend to test on powerpc and ppc64 (perotto.debian.net) and barriere.debian.org
(i386 and amd64).
Adrian
--
.''`. John Paul Adrian Glaubitz
: :' : Debian Developer
`. `' Physicist
`- GPG: 62FF 8A75 84E0 2956 9546 0006 7426 3B37 F5B5 F913
Reply to: