The null buffer list check in Append ran for every batch-job builder,
although only sceAjmBatchJobRunSplit and sceAjmBatchJobControl pass a
caller's buffer lists. It now runs in CheckedAppend, after the same
invalid-parameter checks, for those two and for
sceAjmBatchJobGetStatistics, whose zero-filled result has no source.
The other builders append without it. No job is built or run
differently.
The guest test builds Initialize, ClearContext, SetGaplessDecode,
SetResampleParameters, SetResampleParametersEx and GetResampleInfo jobs
without buffers and runs them, and checks that RunSplit still throws
for a null input or output list with a nonzero count without writing
to the batch.
The RDNA program decode sweeps linearly up to the last s_endpgm past the
furthest direct branch target, so it also decoded words that no path
reaches and threw on any it could not decode. A word directly after
s_branch, s_endpgm or s_code_end that fails to decode is now skipped;
the decode still throws that word's error when a direct branch or
s_call_b64 targets it, and a failure on a word reachable by fall-through
is unchanged. The control-flow graph starts the block after s_branch and
s_endpgm at the next decoded instruction instead of the next word.
Co-Authored-By: Claude Opus 5.5 <[email protected]>
On PS5 firmware 13.60 sceRtcSetTime_t(NULL, -1) returns
SCE_RTC_ERROR_INVALID_VALUE and sceRtcSetTime_t(NULL, 0)
SCE_RTC_ERROR_INVALID_POINTER: negative seconds are rejected before a
null destination under the later SDK rules. The other measured
GetDayOfWeek, SetTime_t and FormatRFC3339 results already matched.
Co-Authored-By: Claude Opus 5.5 <[email protected]>
sched_getparam, sched_getscheduler and sched_setparam are defined under libkernel/Pthread/Posix, so their typed declarations replace the no-argument placeholders in the libkernel header (public_headers requires the declaration in the implementing library's header), and GuestSchedParams.cpp includes the canonical headers instead of declaring these exports and the pthread, getpid and error calls locally.
FreeBSD's pollscan clears POLLOUT whenever POLLHUP is set, for every
descriptor. 64ca6e332 applied that to pipes, but a stream socket whose
peer has closed and whose own sending side is shut down still reported
POLLOUT next to POLLHUP, because Linux's poll sets both. Linux sets
POLLHUP only once both directions are shut, the same condition as
FreeBSD, so poll now clears POLLOUT from any entry that reports POLLHUP
there.
Apply the firmware 13.42 module-probe report in PR #9506: all four DNS calls check initialization first; getters return EINVAL for null outputs or nonzero flags; setters reject flags before reading guest memory. Keep the checked input copy and document EFAULT as host safety, not firmware behavior. Use the current public Net.hpp declarations in the regression.
Rebased onto upstream e7e3e9c198. Ubuntu 24.04, GCC 13.3, Release: actual libc/libkernel/libSceNet libraries and both focused test executables built. guest_sce_net passes. A separate harness linked against those unmodified libraries passes initialization, errno, invalid flags, unreadable-input flag precedence, null clearing and unchanged-output checks. The full guest_net_dns_info test is blocked here: process_vm_readv is denied with EPERM by the execution environment. No full DNS regression pass is claimed. Public-header audit, conventions, test registration and whitespace checks pass.
The resolver still uses host DNS; IPv6 host-server discovery and firmware server-count behavior remain debt. Firmware evidence: https://github.com/boykopovar/AnyPS5/pull/9506#issuecomment-6390574180
AI-assisted: yes
On PS5 firmware 13.60 setpriority(PRIO_PROCESS, 0, v) returns 0 for -25,
-5, 5 and 25, and getpriority keeps returning 0. setpriority keeps its
target checks and returns 0 without storing the value.
Co-Authored-By: Claude Opus 5.5 <[email protected]>
A descriptor view inside an imported range of the cached address space
is served by the space's region and gets no region of its own. When
UploadFinish's join of two such ranges fails at vkAllocateMemory, the
join has already retired the ranges' imports and the lowered ceiling
refuses new ones, so the dissolved space's region falls back to a copy
of the registered range. For a range the CPU cannot write, that copy is
read-only: the dispatch's stores land in a buffer that is never written
back, and WriteBack fails with "write-back range exceeds its GPU owner".
The dissolve that re-resolves the space's imports now marks a region
written by one of the build's views as writable, as UploadPrepare's merge
does for a view with a region of its own, so the copy keeps its reference
bytes and the stores are written back.
Save r12 in the entry walk so TraceImport_nid_no_patch is called with an aligned stack, record only executable Mach-O segments in the module map, and make the test hook check the alignment.
Co-Authored-By: Claude Sonnet 5.5 <[email protected]>
Walk each macOS function import before the guest entry and pass it through TraceImport_nid_no_patch, as Linux and Windows already do. An x86-64 libc build records the call and prints it on SIGABRT.
Both exports threw and were declared without parameters. They follow
LibreSSL 2.5.4, unchanged in 2.7: X509_LOOKUP_file returns the static
80-byte file method, and X509_STORE_add_lookup returns the store's
lookup for a method, or mallocs a 32-byte lookup owned by the store and
pushes it on the store's methods, growing the stack the way sk_insert
does. X509_STORE_free now frees the store's lookups.
The file method's ctrl, which loads PEM files, throws: there is no
certificate model yet. Callbacks of guest-supplied methods throw
instead of being called.
Co-Authored-By: Claude Opus 5.5 <[email protected]>
Both exports threw and were declared without parameters. They follow
LibreSSL 2.5.4, unchanged in 2.7: set_hostflags stores the host
matching flags in the verify parameters' identity, and set1_ip_asc
parses an IPv4 or IPv6 address the way a2i_ipadd does, copies the 4 or
16 bytes into a guest malloc block that replaces the previous address,
and returns 0 without touching it for text the parser rejects. IPv4
parts are read with the guest sscanf, as the module would.
Co-Authored-By: Claude Opus 5.5 <[email protected]>
The three exports threw and were declared without parameters. They now
follow LibreSSL 2.5.4, the version the firmware 13.40 measurements
identify: X509_STORE_new allocates the 136-byte store with an empty
object cache, an empty lookup method stack and 64-byte verify
parameters at depth -1, X509_STORE_free drops a reference and frees the
store with its stacks, verify parameters and their host names, peer
name, email and IP, and X509_STORE_set_flags ORs the flags into the
verify parameters and adds the policy check for any policy flag.
No export adds certificates, CRLs or lookup methods yet, so freeing a
store that holds them, or verify parameters holding policy objects,
throws instead of freeing objects with no model.
Co-Authored-By: Claude Opus 5.5 <[email protected]>
Adrià