September 2026 - gerrit-log - lists.osmocom.org

[XS] Change in osmo-trx[master]: build: pass FFTWF_CFLAGS to arch/common
by Andrei G 08 Sep '26

08 Sep '26

Andrei G has uploaded this change for review. ( https://gerrit.osmocom.org/c/osmo-trx/+/43581?usp=email ) Change subject: build: pass FFTWF_CFLAGS to arch/common ...................................................................... build: pass FFTWF_CFLAGS to arch/common configure looks up fftw3f with pkg-config for the multi-ARFCN build and puts FFTWF_LIBS on the link line of the transceiver, but Transceiver52M/arch/common/Makefile.am, where fft.c lives, never uses FFTWF_CFLAGS. The compile of fft.c therefore only works where fftw3.h is in a default include directory. With fftw from Homebrew on macOS ARM64 the header is in /opt/homebrew/include, which clang does not search by default: arch/common/fft.c:26:10: fatal error: 'fftw3.h' file not found Add FFTWF_CFLAGS to AM_CFLAGS in that directory. No change where the header was already found. Change-Id: I3243ce00bd42a22ff0fbbdd5ff3d00c7441662d0 Signed-off-by: Andrei Gosman <andrei.gosman(a)gmail.com> --- M Transceiver52M/arch/common/Makefile.am 1 file changed, 1 insertion(+), 1 deletion(-) git pull ssh://gerrit.osmocom.org:29418/osmo-trx refs/changes/81/43581/1 diff --git a/Transceiver52M/arch/common/Makefile.am b/Transceiver52M/arch/common/Makefile.am index a27174d..c23d3b4 100644 --- a/Transceiver52M/arch/common/Makefile.am +++ b/Transceiver52M/arch/common/Makefile.am @@ -1,4 +1,4 @@ -AM_CFLAGS = -Wall -std=gnu99 +AM_CFLAGS = -Wall -std=gnu99 $(FFTWF_CFLAGS) noinst_LTLIBRARIES = libarch_common.la -- To view, visit https://gerrit.osmocom.org/c/osmo-trx/+/43581?usp=email To unsubscribe, or for help writing mail filters, visit https://gerrit.osmocom.org/settings?usp=email Gerrit-MessageType: newchange Gerrit-Project: osmo-trx Gerrit-Branch: master Gerrit-Change-Id: I3243ce00bd42a22ff0fbbdd5ff3d00c7441662d0 Gerrit-Change-Number: 43581 Gerrit-PatchSet: 1 Gerrit-Owner: Andrei G <andrei.gosman(a)gmail.com>

1 0

[XS] Change in osmo-trx[master]: build: probe UHD version with PKG_CHECK_EXISTS
by Andrei G 08 Sep '26

08 Sep '26

Andrei G has uploaded this change for review. ( https://gerrit.osmocom.org/c/osmo-trx/+/43580?usp=email ) Change subject: build: probe UHD version with PKG_CHECK_EXISTS ...................................................................... build: probe UHD version with PKG_CHECK_EXISTS configure.ac appends -lboost_thread for libuhd < 4.2, which needs it because of a bug in its own code. The probe is a second PKG_CHECK_MODULES(UHD, uhd < 004.002, ...) reusing the UHD variable prefix of the first one. That never asks pkg-config. PKG_CHECK_MODULES expands _PKG_CONFIG, which opens with if test -n "$$1"; then pkg_cv_[]$1="$$1" so once UHD_CFLAGS and UHD_LIBS are set by the earlier PKG_CHECK_MODULES(UHD, uhd >= 003.011), the second call takes those values as its answer, leaves pkg_failed at no, and runs the action-if-found branch whatever the installed version is. -lboost_thread is therefore always appended. On GNU/Linux that only adds a library that is usually present, so the bug is invisible. On macOS ARM64 with UHD 4.10 from Homebrew there is no such library on the link path and the build stops: ld: library 'boost_thread' not found Use PKG_CHECK_EXISTS, which asks pkg-config and sets no variables. The workaround then applies where it was meant to, on libuhd < 4.2. Change-Id: Icbd8af94f258128ea75c0268e0a240dc423ff69e Signed-off-by: Andrei Gosman <andrei.gosman(a)gmail.com> --- M configure.ac 1 file changed, 2 insertions(+), 3 deletions(-) git pull ssh://gerrit.osmocom.org:29418/osmo-trx refs/changes/80/43580/1 diff --git a/configure.ac b/configure.ac index a384495..0cd388d 100644 --- a/configure.ac +++ b/configure.ac @@ -195,9 +195,8 @@ # code and therefore requires linking against boost_thread. It's missing in # uhd.pc, so work around it here. # https://github.com/EttusResearch/uhd/commit/04a83b6e76beef970854da69ba882d7… - PKG_CHECK_MODULES(UHD, uhd < 004.002, - [LIBS="$LIBS -lboost_thread"], - [] + PKG_CHECK_EXISTS([uhd < 004.002], + [LIBS="$LIBS -lboost_thread"] ) ]) -- To view, visit https://gerrit.osmocom.org/c/osmo-trx/+/43580?usp=email To unsubscribe, or for help writing mail filters, visit https://gerrit.osmocom.org/settings?usp=email Gerrit-MessageType: newchange Gerrit-Project: osmo-trx Gerrit-Branch: master Gerrit-Change-Id: Icbd8af94f258128ea75c0268e0a240dc423ff69e Gerrit-Change-Number: 43580 Gerrit-PatchSet: 1 Gerrit-Owner: Andrei G <andrei.gosman(a)gmail.com>

1 0

[XS] Change in osmo-trx[master]: build: require C++17 for the UHD 4.x headers
by Andrei G 08 Sep '26

08 Sep '26

Andrei G has uploaded this change for review. ( https://gerrit.osmocom.org/c/osmo-trx/+/43579?usp=email ) Change subject: build: require C++17 for the UHD 4.x headers ...................................................................... build: require C++17 for the UHD 4.x headers configure asks for C++11 through AX_CXX_COMPILE_STDCXX_11. The public headers of UHD 4.10 use std::optional, std::is_same_v and std::is_arithmetic_v, none of which exist in C++11, so UHDDevice.cpp fails to compile against them: uhd/utils/cast.hpp:70:15: error: no template named 'is_same_v' in namespace 'std' uhd/rfnoc/actions.hpp:104:15: error: no template named 'optional' in namespace 'std' UHD 4.10 builds its own sources as C++20, and a consumer of its public headers needs at least C++17. Ask for C++17 with AX_CXX_COMPILE_STDCXX, the same macro already shipped in config/. The osmo-trx sources compile unchanged under the newer standard. Seen with UHD 4.10.0.0 from Homebrew on macOS ARM64. The version, not the platform, is what decides this: any distribution carrying UHD 4.10 reaches the same headers. Change-Id: If8d80f0708f3856f36d56e637ad8d57f3d3d1bad Signed-off-by: Andrei Gosman <andrei.gosman(a)gmail.com> --- M configure.ac 1 file changed, 1 insertion(+), 1 deletion(-) git pull ssh://gerrit.osmocom.org:29418/osmo-trx refs/changes/79/43579/1 diff --git a/configure.ac b/configure.ac index a384495..d7795b3 100644 --- a/configure.ac +++ b/configure.ac @@ -46,7 +46,7 @@ AM_PROG_AS AC_PROG_CC AC_PROG_CXX -AX_CXX_COMPILE_STDCXX_11 +AX_CXX_COMPILE_STDCXX([17], [ext], [mandatory]) AC_PROG_LN_S AC_PROG_MAKE_SET AC_PROG_INSTALL -- To view, visit https://gerrit.osmocom.org/c/osmo-trx/+/43579?usp=email To unsubscribe, or for help writing mail filters, visit https://gerrit.osmocom.org/settings?usp=email Gerrit-MessageType: newchange Gerrit-Project: osmo-trx Gerrit-Branch: master Gerrit-Change-Id: If8d80f0708f3856f36d56e637ad8d57f3d3d1bad Gerrit-Change-Number: 43579 Gerrit-PatchSet: 1 Gerrit-Owner: Andrei G <andrei.gosman(a)gmail.com>

1 0

[XS] Change in osmo-hlr[master]: gsupclient: add LIBOSMOGSM_LIBS to LIBADD
by Andrei G 08 Sep '26

08 Sep '26

Andrei G has uploaded this change for review. ( https://gerrit.osmocom.org/c/osmo-hlr/+/43578?usp=email ) Change subject: gsupclient: add LIBOSMOGSM_LIBS to LIBADD ...................................................................... gsupclient: add LIBOSMOGSM_LIBS to LIBADD libosmo-gsup-client calls osmo_gsup_*, osmo_oap_client_* and osmo_imsi_str_valid(), all exported by libosmogsm, but libosmo_gsup_client_la_LIBADD in src/gsupclient/Makefile.am lists only talloc, libosmocore and libosmoabis. GNU ld accepts undefined symbols in a shared library and defers them to whatever the executable happens to pull in, so the missing dependency is invisible on Linux. Darwin's ld64 refuses at link time and the library fails to build with "symbol(s) not found for architecture arm64". The same Makefile.am already passes -no-undefined in LDFLAGS, which is the promise that the library resolves its own symbols. Adding LIBOSMOGSM_LIBS is what makes that promise true. With the fix, otool -L on the resulting libosmo-gsup-client.dylib lists libosmogsm among its dependencies, and the osmo_gsup_* references resolve through it. Change-Id: I413ff47aa1e8034ea86c61134d3d6784942b0e3b Signed-off-by: Andrei Gosman <andrei.gosman(a)gmail.com> --- M src/gsupclient/Makefile.am 1 file changed, 1 insertion(+), 1 deletion(-) git pull ssh://gerrit.osmocom.org:29418/osmo-hlr refs/changes/78/43578/1 diff --git a/src/gsupclient/Makefile.am b/src/gsupclient/Makefile.am index 7611f6e..89533f4 100644 --- a/src/gsupclient/Makefile.am +++ b/src/gsupclient/Makefile.am @@ -16,7 +16,7 @@ $(NULL) libosmo_gsup_client_la_LDFLAGS = -version-info $(LIBVERSION) -no-undefined -libosmo_gsup_client_la_LIBADD = $(TALLOC_LIBS) $(LIBOSMOCORE_LIBS) $(LIBOSMOABIS_LIBS) +libosmo_gsup_client_la_LIBADD = $(TALLOC_LIBS) $(LIBOSMOCORE_LIBS) $(LIBOSMOGSM_LIBS) $(LIBOSMOABIS_LIBS) noinst_PROGRAMS = gsup-test-client -- To view, visit https://gerrit.osmocom.org/c/osmo-hlr/+/43578?usp=email To unsubscribe, or for help writing mail filters, visit https://gerrit.osmocom.org/settings?usp=email Gerrit-MessageType: newchange Gerrit-Project: osmo-hlr Gerrit-Branch: master Gerrit-Change-Id: I413ff47aa1e8034ea86c61134d3d6784942b0e3b Gerrit-Change-Number: 43578 Gerrit-PatchSet: 1 Gerrit-Owner: Andrei G <andrei.gosman(a)gmail.com>

1 0

[XS] Change in osmo-mgw[master]: mgcp_client: include errno.h, not asm-generic
by Andrei G 08 Sep '26

08 Sep '26

Andrei G has uploaded this change for review. ( https://gerrit.osmocom.org/c/osmo-mgw/+/43577?usp=email ) Change subject: mgcp_client: include errno.h, not asm-generic ...................................................................... mgcp_client: include errno.h, not asm-generic src/libosmo-mgcp-client/mgcp_client_pool.c includes <asm-generic/errno.h> for ECONNABORTED and EINVAL. That is a Linux kernel UAPI header. It exists on Linux distributions and nowhere else, so the build stops on Darwin: mgcp_client_pool.c:21:10: fatal error: 'asm-generic/errno.h' file not found Both values come from <errno.h> on every platform, and on Linux glibc's <errno.h> reaches the same UAPI definitions, so including the kernel header directly was never needed there either. No functional change on any platform. Change-Id: Ie3fc368c04c29aea824002b3bf9839e410e7f045 Signed-off-by: Andrei Gosman <andrei.gosman(a)gmail.com> --- M src/libosmo-mgcp-client/mgcp_client_pool.c 1 file changed, 1 insertion(+), 1 deletion(-) git pull ssh://gerrit.osmocom.org:29418/osmo-mgw refs/changes/77/43577/1 diff --git a/src/libosmo-mgcp-client/mgcp_client_pool.c b/src/libosmo-mgcp-client/mgcp_client_pool.c index 61fb134..42d775e 100644 --- a/src/libosmo-mgcp-client/mgcp_client_pool.c +++ b/src/libosmo-mgcp-client/mgcp_client_pool.c @@ -18,7 +18,7 @@ * */ -#include <asm-generic/errno.h> +#include <errno.h> #include <osmocom/mgcp_client/mgcp_client.h> #include <osmocom/mgcp_client/mgcp_client_internal.h> #include <osmocom/mgcp_client/mgcp_client_pool_internal.h> -- To view, visit https://gerrit.osmocom.org/c/osmo-mgw/+/43577?usp=email To unsubscribe, or for help writing mail filters, visit https://gerrit.osmocom.org/settings?usp=email Gerrit-MessageType: newchange Gerrit-Project: osmo-mgw Gerrit-Branch: master Gerrit-Change-Id: Ie3fc368c04c29aea824002b3bf9839e410e7f045 Gerrit-Change-Number: 43577 Gerrit-PatchSet: 1 Gerrit-Owner: Andrei G <andrei.gosman(a)gmail.com>

1 0

[XS] Change in osmo-mgw[master]: mgcp_client: generate mgcp_common.h with printf
by Andrei G 08 Sep '26

08 Sep '26

Andrei G has uploaded a new patch set (#2). ( https://gerrit.osmocom.org/c/osmo-mgw/+/43576?usp=email ) Change subject: mgcp_client: generate mgcp_common.h with printf ...................................................................... mgcp_client: generate mgcp_common.h with printf The rule in include/osmocom/mgcp_client/Makefile.am that copies <osmocom/mgcp/mgcp_common.h> into the mgcp_client include directory writes its banner with "echo -e". The -e option is a bash extension. Shells whose echo follows the XSI convention expand the backslash escapes on their own and do not take options, so -e is passed through as the first argument to print. The generated header then starts with -e /* and every file that includes it fails to compile: mgcp_common.h:1:2: error: unknown type name 'e' Measured on macOS, where /bin/sh is bash invoked as sh: the escapes are expanded as intended, but the leading "-e " is printed literally. Any shell with an XSI-style echo, dash among them, behaves the same way. The build only works where /bin/sh accepts -e. Replace it with printf and one argument per line. printf is POSIX, takes no such option, and produces the same seven lines on every shell. Change-Id: Ibc547513745104212469ccddbc2e636c985cffe5 Signed-off-by: Andrei Gosman <andrei.gosman(a)gmail.com> --- M include/osmocom/mgcp_client/Makefile.am 1 file changed, 1 insertion(+), 1 deletion(-) git pull ssh://gerrit.osmocom.org:29418/osmo-mgw refs/changes/76/43576/2 -- To view, visit https://gerrit.osmocom.org/c/osmo-mgw/+/43576?usp=email To unsubscribe, or for help writing mail filters, visit https://gerrit.osmocom.org/settings?usp=email Gerrit-MessageType: newpatchset Gerrit-Project: osmo-mgw Gerrit-Branch: master Gerrit-Change-Id: Ibc547513745104212469ccddbc2e636c985cffe5 Gerrit-Change-Number: 43576 Gerrit-PatchSet: 2 Gerrit-Owner: Andrei G <andrei.gosman(a)gmail.com>

1 0

[XS] Change in osmo-mgw[master]: mgcp_client: generate mgcp_common.h banner with printf
by Andrei G 08 Sep '26

08 Sep '26

Andrei G has uploaded this change for review. ( https://gerrit.osmocom.org/c/osmo-mgw/+/43576?usp=email ) Change subject: mgcp_client: generate mgcp_common.h banner with printf ...................................................................... mgcp_client: generate mgcp_common.h banner with printf The rule in include/osmocom/mgcp_client/Makefile.am that copies <osmocom/mgcp/mgcp_common.h> into the mgcp_client include directory writes its banner with "echo -e". The -e option is a bash extension. Shells whose echo follows the XSI convention expand the backslash escapes on their own and do not take options, so -e is passed through as the first argument to print. The generated header then starts with -e /* and every file that includes it fails to compile: mgcp_common.h:1:2: error: unknown type name 'e' Measured on macOS, where /bin/sh is bash invoked as sh: the escapes are expanded as intended, but the leading "-e " is printed literally. Any shell with an XSI-style echo, dash among them, behaves the same way. The build only works where /bin/sh accepts -e. Replace it with printf and one argument per line. printf is POSIX, takes no such option, and produces the same seven lines on every shell. Change-Id: Ibc547513745104212469ccddbc2e636c985cffe5 Signed-off-by: Andrei Gosman <andrei.gosman(a)gmail.com> --- M include/osmocom/mgcp_client/Makefile.am 1 file changed, 1 insertion(+), 1 deletion(-) git pull ssh://gerrit.osmocom.org:29418/osmo-mgw refs/changes/76/43576/1 diff --git a/include/osmocom/mgcp_client/Makefile.am b/include/osmocom/mgcp_client/Makefile.am index 7415142..5374dc5 100644 --- a/include/osmocom/mgcp_client/Makefile.am +++ b/include/osmocom/mgcp_client/Makefile.am @@ -9,7 +9,7 @@ $(NULL) mgcp_common.h: $(top_srcdir)/include/osmocom/mgcp/mgcp_common.h - echo -e "/*\n\n DO NOT EDIT THIS FILE!\n THIS IS OVERWRITTEN DURING BUILD\n This is an automatic copy of <osmocom/mgcp/mgcp_common.h>\n\n */" > mgcp_common.h + printf '%s\n' "/*" "" " DO NOT EDIT THIS FILE!" " THIS IS OVERWRITTEN DURING BUILD" " This is an automatic copy of <osmocom/mgcp/mgcp_common.h>" "" " */" > mgcp_common.h cat $(top_srcdir)/include/osmocom/mgcp/mgcp_common.h >> mgcp_common.h version.h: version.h.tpl -- To view, visit https://gerrit.osmocom.org/c/osmo-mgw/+/43576?usp=email To unsubscribe, or for help writing mail filters, visit https://gerrit.osmocom.org/settings?usp=email Gerrit-MessageType: newchange Gerrit-Project: osmo-mgw Gerrit-Branch: master Gerrit-Change-Id: Ibc547513745104212469ccddbc2e636c985cffe5 Gerrit-Change-Number: 43576 Gerrit-PatchSet: 1 Gerrit-Owner: Andrei G <andrei.gosman(a)gmail.com>

1 0

[M] Change in libosmocore[master]: core/socket: sockaddr_cmp: compare fields
by Andrei G 08 Sep '26

08 Sep '26

Hello Jenkins Builder, I'd like you to reexamine a change. Please visit https://gerrit.osmocom.org/c/libosmocore/+/43575?usp=email to look at the new patch set (#3). The following approvals got outdated and were removed: Verified+1 by Jenkins Builder Change subject: core/socket: sockaddr_cmp: compare fields ...................................................................... core/socket: sockaddr_cmp: compare fields osmo_sockaddr_cmp() decides whether two addresses are equal with a memcmp() over the whole struct sockaddr_in or sockaddr_in6. On Linux those structures hold only family, port, address and padding, so the comparison is exact. On Darwin and the BSDs the first byte is sin_len / sin6_len. The kernel fills it in on recvfrom() and accept(), while an address the application built from configuration leaves it zero. The same peer then compares as two different addresses. Every static NS-VC in gprs_ns2 breaks on this. The answer to the first NS-RESET arrives from the configured remote, gprs_ns2_udp.c cannot match it to the NS-VC ("Ignoring NS RESET ACK from newconnection for non-existing NS-VC", gprs_ns2.c:1066) and the link never leaves RESET. Observed with osmo-pcu against osmo-sgsn on Darwin loopback: the SGSN side, which learned the peer from the packet, goes to BLOCKED and then loses every NS-ALIVE-ACK, while the PCU side stays in RESET. Dynamic NS-VCs are unaffected because their remote address is a copy of what recvfrom() returned. Compare the fields instead of the bytes: port and address for AF_INET, port, flow info, address and scope id for AF_INET6, in the order memcmp() visited them, so the ordering the function gives to sorted users does not change. The default branch keeps its memcmp() over the full osmo_sockaddr. Linux behaviour is unchanged. Add tests/sockaddr_cmp covering the two cases that regressed and the ordering the fix has to preserve: an IPv4 and an IPv6 pair equal in port and address but differing in the length byte, both expected to compare equal, and ordering by port, by address, by flow info and by scope id, each checked in both directions. Against the previous implementation the two length-byte cases return -16 and -28 instead of 0. The length byte only exists where the sockaddr carries one, so configure.ac gains an AC_CHECK_MEMBER for struct sockaddr_in.sin_len, in the style of the existing s6_addr32 check. Where the member is absent the two addresses are identical and the expectation is the same, so the test and its .ok output are the same on every platform. Note that sin6_flowinfo and sin6_scope_id are compared byte by byte, which is not numeric order on a little endian host. That is unchanged from the memcmp() this replaces. The test uses values differing in one byte only, where both orders agree. Change-Id: I75fc62a92f546a906dbb1440984d62f962579255 Signed-off-by: Andrei Gosman <andrei.gosman(a)gmail.com> --- M configure.ac M src/core/socket.c M tests/Makefile.am A tests/sockaddr_cmp/sockaddr_cmp_test.c A tests/sockaddr_cmp/sockaddr_cmp_test.ok M tests/testsuite.at 6 files changed, 201 insertions(+), 4 deletions(-) git pull ssh://gerrit.osmocom.org:29418/libosmocore refs/changes/75/43575/3 -- To view, visit https://gerrit.osmocom.org/c/libosmocore/+/43575?usp=email To unsubscribe, or for help writing mail filters, visit https://gerrit.osmocom.org/settings?usp=email Gerrit-MessageType: newpatchset Gerrit-Project: libosmocore Gerrit-Branch: master Gerrit-Change-Id: I75fc62a92f546a906dbb1440984d62f962579255 Gerrit-Change-Number: 43575 Gerrit-PatchSet: 3 Gerrit-Owner: Andrei G <andrei.gosman(a)gmail.com> Gerrit-Reviewer: Jenkins Builder

1 0

[S] Change in libosmocore[master]: core/socket: sockaddr_cmp: compare fields
by Andrei G 08 Sep '26

08 Sep '26

Andrei G has uploaded a new patch set (#2). ( https://gerrit.osmocom.org/c/libosmocore/+/43575?usp=email ) Change subject: core/socket: sockaddr_cmp: compare fields ...................................................................... core/socket: sockaddr_cmp: compare fields osmo_sockaddr_cmp() decides whether two addresses are equal with a memcmp() over the whole struct sockaddr_in or sockaddr_in6. On Linux those structures hold only family, port, address and padding, so the comparison is exact. On Darwin and the BSDs the first byte is sin_len / sin6_len. The kernel fills it in on recvfrom() and accept(), while an address the application built from configuration leaves it zero. The same peer then compares as two different addresses. Every static NS-VC in gprs_ns2 breaks on this. The answer to the first NS-RESET arrives from the configured remote, gprs_ns2_udp.c cannot match it to the NS-VC ("Ignoring NS RESET ACK from newconnection for non-existing NS-VC", gprs_ns2.c:1066) and the link never leaves RESET. Observed with osmo-pcu against osmo-sgsn on Darwin loopback: the SGSN side, which learned the peer from the packet, goes to BLOCKED and then loses every NS-ALIVE-ACK, while the PCU side stays in RESET. Dynamic NS-VCs are unaffected because their remote address is a copy of what recvfrom() returned. Compare the fields instead of the bytes: port and address for AF_INET, port, flow info, address and scope id for AF_INET6, in the order memcmp() visited them, so the ordering the function gives to sorted users does not change. The default branch keeps its memcmp() over the full osmo_sockaddr. Linux behaviour is unchanged. Change-Id: I75fc62a92f546a906dbb1440984d62f962579255 Signed-off-by: Andrei Gosman <andrei.gosman(a)gmail.com> --- M src/core/socket.c 1 file changed, 30 insertions(+), 4 deletions(-) git pull ssh://gerrit.osmocom.org:29418/libosmocore refs/changes/75/43575/2 -- To view, visit https://gerrit.osmocom.org/c/libosmocore/+/43575?usp=email To unsubscribe, or for help writing mail filters, visit https://gerrit.osmocom.org/settings?usp=email Gerrit-MessageType: newpatchset Gerrit-Project: libosmocore Gerrit-Branch: master Gerrit-Change-Id: I75fc62a92f546a906dbb1440984d62f962579255 Gerrit-Change-Number: 43575 Gerrit-PatchSet: 2 Gerrit-Owner: Andrei G <andrei.gosman(a)gmail.com>

1 0

[S] Change in libosmocore[master]: core/socket: sockaddr_cmp: compare fields, not bytes
by Andrei G 08 Sep '26

08 Sep '26

Andrei G has uploaded this change for review. ( https://gerrit.osmocom.org/c/libosmocore/+/43575?usp=email ) Change subject: core/socket: sockaddr_cmp: compare fields, not bytes ...................................................................... core/socket: sockaddr_cmp: compare fields, not bytes osmo_sockaddr_cmp() decides whether two addresses are equal with a memcmp() over the whole struct sockaddr_in or sockaddr_in6. On Linux those structures hold only family, port, address and padding, so the comparison is exact. On Darwin and the BSDs the first byte is sin_len / sin6_len. The kernel fills it in on recvfrom() and accept(), while an address the application built from configuration leaves it zero. The same peer then compares as two different addresses. Every static NS-VC in gprs_ns2 breaks on this. The answer to the first NS-RESET arrives from the configured remote, gprs_ns2_udp.c cannot match it to the NS-VC ("Ignoring NS RESET ACK from newconnection for non-existing NS-VC", gprs_ns2.c:1066) and the link never leaves RESET. Observed with osmo-pcu against osmo-sgsn on Darwin loopback: the SGSN side, which learned the peer from the packet, goes to BLOCKED and then loses every NS-ALIVE-ACK, while the PCU side stays in RESET. Dynamic NS-VCs are unaffected because their remote address is a copy of what recvfrom() returned. Compare the fields instead of the bytes: port and address for AF_INET, port, flow info, address and scope id for AF_INET6, in the order memcmp() visited them, so the ordering the function gives to sorted users does not change. The default branch keeps its memcmp() over the full osmo_sockaddr. Linux behaviour is unchanged. Change-Id: I75fc62a92f546a906dbb1440984d62f962579255 Signed-off-by: Andrei Gosman <andrei.gosman(a)gmail.com> --- M src/core/socket.c 1 file changed, 30 insertions(+), 4 deletions(-) git pull ssh://gerrit.osmocom.org:29418/libosmocore refs/changes/75/43575/1 diff --git a/src/core/socket.c b/src/core/socket.c index 3a2d7b7..724d8ef 100644 --- a/src/core/socket.c +++ b/src/core/socket.c @@ -2599,11 +2599,37 @@ return OSMO_CMP(a->u.sa.sa_family, b->u.sa.sa_family); } + /* Compare the address fields, not the raw bytes of the structure. + * On the BSDs and Darwin the sockaddr carries a length byte (sin_len, + * sin6_len) that the kernel fills in on recvfrom()/accept() and that + * is zero in an address the application built itself; a memcmp() over + * the whole structure then reports two equal addresses as different. + * The order of the comparisons keeps the ordering memcmp() gave. */ switch (a->u.sa.sa_family) { - case AF_INET: - return memcmp(&a->u.sin, &b->u.sin, sizeof(struct sockaddr_in)); - case AF_INET6: - return memcmp(&a->u.sin6, &b->u.sin6, sizeof(struct sockaddr_in6)); + case AF_INET: { + int rc = memcmp(&a->u.sin.sin_port, &b->u.sin.sin_port, + sizeof(a->u.sin.sin_port)); + if (rc) + return rc; + return memcmp(&a->u.sin.sin_addr, &b->u.sin.sin_addr, + sizeof(a->u.sin.sin_addr)); + } + case AF_INET6: { + int rc = memcmp(&a->u.sin6.sin6_port, &b->u.sin6.sin6_port, + sizeof(a->u.sin6.sin6_port)); + if (rc) + return rc; + rc = memcmp(&a->u.sin6.sin6_flowinfo, &b->u.sin6.sin6_flowinfo, + sizeof(a->u.sin6.sin6_flowinfo)); + if (rc) + return rc; + rc = memcmp(&a->u.sin6.sin6_addr, &b->u.sin6.sin6_addr, + sizeof(a->u.sin6.sin6_addr)); + if (rc) + return rc; + return memcmp(&a->u.sin6.sin6_scope_id, &b->u.sin6.sin6_scope_id, + sizeof(a->u.sin6.sin6_scope_id)); + } default: /* fallback to memcmp for remaining AF over the full osmo_sockaddr length */ return memcmp(a, b, sizeof(struct osmo_sockaddr)); -- To view, visit https://gerrit.osmocom.org/c/libosmocore/+/43575?usp=email To unsubscribe, or for help writing mail filters, visit https://gerrit.osmocom.org/settings?usp=email Gerrit-MessageType: newchange Gerrit-Project: libosmocore Gerrit-Branch: master Gerrit-Change-Id: I75fc62a92f546a906dbb1440984d62f962579255 Gerrit-Change-Number: 43575 Gerrit-PatchSet: 1 Gerrit-Owner: Andrei G <andrei.gosman(a)gmail.com>

1 0

2026

2025

2024

2023

2022

gerrit-log September 2026