Control: reassign -1 libgcrypt20 1.11.2-3 On Sun, Jan 18, 2026 at 07:28:22AM +0100, Andreas Metzler wrote: > Control: reassign 1125774 ndpi 4.2-2.1 > > On 2026-01-17 Emanuele Rocca <[email protected]> wrote: > > On 2026-01-17 06:15, Andreas Metzler wrote: > > > configure:22126: checking for gcry_cipher_checktag in -lgcrypt > > > configure:22155: gcc -o conftest -W -Wall -Wno-unused-parameter > > > -Wno-unused-function -g -O2 -Werror=implicit-function-declaration > > > -ffile-prefix-map=/home/ametzler/NDPI/ndpi-4.2=. -fstack-protector-strong > > > -fstack-clash-protection -Wformat -Werror=format-security > > > -mbranch-protection=standard -I/usr/include/json-c -Wdate-time > > > -D_FORTIFY_SOURCE=2 -Wl,-z,relro -Wl,-z,now -mbranch-protection=standard > > > conftest.c -lgcrypt -ljson-c -ljson-c >&5 > > > /usr/lib/gcc/aarch64-linux-gnu/15/../../../aarch64-linux-gnu/libgcrypt.so: > > > warning: GCS is required by -z gcs, but this shared library lacks the > > > necessary property note. The dynamic loader might not enable GCS or > > > refuse to load the program unless all the shared library dependencies > > > have the GCS marking. > > > configure:22155: $? = 0 > > > configure: failed program was: > > > [...] > > > configure:22167: result: no > > > > > > However, taking the snippped "failed program" and running the quoted > > > "gcc -o conftest -W ..." on it yields the warning but exits > > > successfully. (exit code 0). > > > Thanks for your analysis, you're right to be very puzzled. > > > What's happening is that configure.ac uses AC_LANG_WERROR, which means > > compiler warnings are considered errors: > > https://sources.debian.org/src/ndpi/4.2-2.1/configure.ac#L45 > > > That's why suppressing the GCS warning fixes the build, as Adrian found. > > Ah, thank you. > > Let's reassign back to ndpi then.
Why does ld complain about libgcrypt not having GCS enabled (as said already during the build of libgcrypt20)? This is the root cause, that should be fixed unless there is a good reason not to do so. Emanuele might be better qualified to understand why this happens. > cu Andreas cu Adrian
