Hide Forgot
Description of problem: Several applications using p11-kit-proxy module crash when softhsm is installed. Firefox always crashes on quit and opensc-tools may reproduce a similar crash while listing details from that module. Version-Release number of selected component (if applicable): 2.3.0-2.fc27, 2.3.0-3.fc28.1 How reproducible: Always. Steps to Reproduce: opensc tools way: 1. Install softhsm and opensc 2. Run: $ pkcs11-tool -L --module /usr/lib64/p11-kit-proxy.so Firefox way: 1. Install softhsm and firefox 2. Configure NSS to use p11-kit-proxy: # alternatives --install /usr/lib64/libnssckbi.so \ libnssckbi.so.x86_64 /usr/lib64/p11-kit-proxy.so 50 3. Run Firefox (quit Firefox if already opened first) and quit Firefox. Actual results: pkcs11-tool crashes with the command above. firefox crashes on quit (no abrt but breakpad crash report window instead). Removing softhsm works around this issue and p11-kit-proxy works without crashes. Expected results: It shouldn't crash.
I tried to reproduce it, but couldn't. Could you provide more information about your configuration: - NSS and p11-kit package versions - How softhsm is configured (do you have a token initialized and any certificates/keys in it?) (In reply to Francisco de la Peña from comment #0) > 2. Configure NSS to use p11-kit-proxy: > # alternatives --install /usr/lib64/libnssckbi.so \ > libnssckbi.so.x86_64 /usr/lib64/p11-kit-proxy.so 50 This is to explicitly override the default installation, that registers p11-kit-trust.so with priority 30, right?
Closing as duplicate as it is seen to be related. *** This bug has been marked as a duplicate of bug 1607635 ***
> This is to explicitly override the default installation, that registers > p11-kit-trust.so with priority 30, right? Yes, it is. Also happens here with the upcoming F29 crypto policies approach. I didn't configure softhsm, so it should use defaults here. Tested with latest F28 downstream packages, still reproduceable here. > Closing as duplicate as it is seen to be related. Indeed, I can confirm the backtrace is the same there as in bug 1607635, +1.