Repository navigation
Conversation
The public SDK's libkernel, libkernel_sys and libkernel_web stub tables declare _sigintr, the sigset_t that FreeBSD's libc shares between signal and siginterrupt, and no built library provided it (boykopovar#8846). libkernel now exports the 16-byte object, zero at start, and signal records a handler with SA_RESTART unless its signal is in the set, as lib/libc/gen/signal.c does. siginterrupt, which fills the set, is a libSceLibcInternal export; libc does not link libkernel and cannot reach the dispositions, so it stays out and a debt entry for signal says so. Co-Authored-By: Claude Fable 5.1 <noreply@anthropic.com>
|
Other open pull requests touch the same code. Merging one will break the other, or both implement the same thing. If this one builds on them, list them in Depends on.
Some comparisons could not be completed because a pull request conflicts with the base branch. Review these shared files manually before concluding that there is no overlap.
|
|
Manual review of the overlap warning: #3986 changes the signal masks, |
|
Thanks for checking the overlaps by hand. Agreed on the hunks: this PR's production change is the one line in |
This branch has not been deployed
2 participants
What
libkernel.prxexports_sigintr(nQVWJEGHObc), thesigset_tthat FreeBSD's libc shares betweensignalandsiginterrupt:signalgives a handlerSA_RESTARTunless its signal is in the set (lib/libc/gen/signal.c, 11.4). The public SDK's libkernel, libkernel_sys and libkernel_web stub tables declare the object, and no built library provided it, so a registry importing it was reported absent (#8846). The object is 16 bytes (fouruint32, the SDK'ssys/_sigset.h), zero at start, andsignalnow reads it under the registration lock when it records a handler.Not included:
siginterrupt, which fills the set. On the console it is a libSceLibcInternal export (the SDK'slibSceLibcInternal.cdeclares it; the three kernel tables do not), which AnyPS5 serves fromlibc.prx, andlibc.prxdoes not linklibkernel.prx(the dependency runs the other way), so it cannot reach the dispositions. A guest that writes the set itself gets FreeBSD's behaviour fromsignal; the function needs a decision on how libc reaches libkernel's signal state and is left for a follow-up. A debt entry forsignalsays so. Whether the two kernel variants should provide the object as well is the policy question of #8853.Tested
Windows 11, MinGW GCC 15.2.0, Release,
BUILD_TESTING=ON, onmain05f9995:guest_signals, extended: the set starts zero;signalrecordsSA_RESTART; with bit 29 set (signal 30) it records flags 0 and keeps the handler; signal 31 still getsSA_RESTART; bit 127 (signal 128) clears it the same way; with the bits clearedSA_RESTARTcomes back.guest_signalspasses (1 of 1).signalchange alone restored tomain(object and test kept) the test fails at the first flags-0 check; restored, it passes. It exits with0xc0000409(theRequireabort) on the reverted build and passes again on the restored one.tools/import_audit.pyon libkernel: public SDK _sigintr data object has no built provider #8846's one-line registry against the built libraries:implemented 1, absent 0, exit 0 (the issue measuredabsent 1).python tools/check_conventions.py --base origin/main --head HEADpasses.sigaction, a few lines below the changed line insignal, and feat(libkernel): export the underscore entry points of the socket, directory and signal calls #8700 adds entry points aftersigactionandsigsuspend; whichever merges second rebases a few lines. Neither touches_sigintr.Checklist
main; no other open PR implements the same functionsNotImplemented_nid_no_patch); silent stubs are listed under Silent stubs in their library's technical debt file.mdfiles or images added to the repository (attach them to this PR)_sigintrimport of libkernel: public SDK _sigintr data object has no built provider #8846;siginterruptstays open there)🤖 Generated with Claude Code