Says ASAN: Direct leak of 524 byte(s) in 1 object(s) allocated from: #0 0x7f708f0a67cf in __interceptor_malloc ../../../../gcc/libsanitizer/asan/asan_malloc_linux.cpp:145 #1 0x7f707d94bf9e in CRYPTO_malloc crypto/mem.c:196 #2 0x7f707d7bd248 in asn1_item_flags_i2d crypto/asn1/tasn_enc.c:65 #3 0x7f707d7bd1b7 in ASN1_item_i2d crypto/asn1/tasn_enc.c:45 #4 0x7f707d85b7be in i2d_DHparams crypto/dh/dh_asn1.c:54 #5 0x7f7075a82223 in q_i2d_DHparams(dh_st*, unsigned char**) qsslsocket_openssl_symbols.cpp:435 #6 0x7f7075a82223 in QTlsBackendOpenSSL::dhParametersFromPem(QByteArray const&, QByteArray*) const qssldiffiehellmanparameters_openssl.cpp:139 #7 0x7f708ca9b588 in QSslDiffieHellmanParametersPrivate::initFromPem(QByteArray const&) qssldiffiehellmanparameters.cpp:285 #8 0x7f708ca9b588 in QSslDiffieHellmanParameters::fromEncoded(QByteArray const&, QSsl::EncodingFormat) qssldiffiehellmanparameters.cpp:94 #9 0x55fd8a545ebe in tst_QSslDiffieHellmanParameters::constructionPEM() tst_qssldiffiehellmanparameters.cpp:98 [...] The pointer returned in the out-parameter of a i2d_DHparams() call is supposed to be OPENSSL_free()ed by the user (this is not at all obvious from the docs¹, but an SO answer² indicates that's how it should be (as well as asan stopping from complaining with this patch applied)). ¹ https://www.openssl.org/docs/man3.1/man3/i2d_DHparams.html ² https://stackoverflow.com/a/53563669. Amends 2cf63c71ebe139890526057dcc51b24ea6df6c30. [ChangeLog][QtNetwork][SSL] Fixed a memory leak in parsing of PEM-encoded Diffie-Hellman parameters. Pick-to: 6.6 6.5 6.2 5.15 Change-Id: I9ed4a26c4676db1c0d54a1945a4fb5014ce568cd Reviewed-by: Mårten Nordheim <marten.nordheim@qt.io> Reviewed-by: Timur Pocheptsov <timur.pocheptsov@qt.io>