fix: comprehensive boot hardening — crashes, warnings, sensors, bare-metal PS/2

- firmware-loader: handle missing INIT_NOTIFY gracefully (Option<RawFd>)
- driver-params: suppress ENODEV (19) on missing driver-manager scheme
- compositor: add wl_seat.name + pointer capabilities for Qt6 compat
- greeter: use redox QPA (libqredox.so) instead of broken Qt6 Wayland
- greeter: reduce DRM wait 10s→2s, kded6 offscreen QPA to avoid crash
- thermald: add CPU die temperature via MSR IA32_THERM_STATUS (Linux coretemp)
- pcid: diagnostic MCFG warning with Q35 guidance, address validation
- dhcpd: auto-interface detection (P0-dhcpd-auto-iface.patch wired)
- procmgr: SIGCHLD EPERM→debug (kernel limitation, not a bug)
- ps2d LED: warn→debug on modern hw without real PS/2 controller
- ps2d mouse: 10×1s→3×250ms retry, fast-fail on unknown response
- i2c RON: handle empty response from i2cd when no adapters present
- base recipe: wire 6 new/improved patches
- config: remove INIT_SKIP, enable all 14 previously-suppressed daemons
This commit is contained in:
2026-05-06 11:16:18 +01:00
parent 1cdb6d4d99
commit ff5a132a9d
16 changed files with 441 additions and 24 deletions
@@ -0,0 +1,30 @@
--- a/drivers/usb/ucsid/src/main.rs
+++ b/drivers/usb/ucsid/src/main.rs
@@ -785,7 +785,11 @@
.context("failed to read I2C control response")?;
buffer.truncate(count);
let text = std::str::from_utf8(&buffer).context("I2C control response was not UTF-8")?;
- ron::from_str(text).context("failed to decode I2C control response")
+ let trimmed = text.trim();
+ if trimmed.is_empty() {
+ return Ok(I2cControlResponse::AdapterList(Vec::new()));
+ }
+ ron::from_str(trimmed).context("failed to decode I2C control response")
}
fn read_symbol_id(path: &Path) -> Result<Option<String>> {
--- a/drivers/gpio/i2c-gpio-expanderd/src/main.rs
+++ b/drivers/gpio/i2c-gpio-expanderd/src/main.rs
@@ -429,7 +429,11 @@
.context("failed to read I2C control response")?;
buffer.truncate(count);
let text = std::str::from_utf8(&buffer).context("I2C control response was not UTF-8")?;
- ron::from_str(text).context("failed to decode I2C control response")
+ let trimmed = text.trim();
+ if trimmed.is_empty() {
+ return Ok(I2cControlResponse::AdapterList(Vec::new()));
+ }
+ ron::from_str(trimmed).context("failed to decode I2C control response")
}
fn eisa_id_from_integer(integer: u64) -> String {
@@ -0,0 +1,148 @@
--- a/drivers/pcid/src/cfg_access/mod.rs
+++ b/drivers/pcid/src/cfg_access/mod.rs
@@ -152,20 +152,22 @@
fn with<T>(
f: impl FnOnce(PcieAllocs<'_>, Vec<InterruptMap>, [u32; 4]) -> io::Result<T>,
) -> io::Result<T> {
- let table_dir = fs::read_dir("/scheme/acpi/tables")?;
+ let table_dir = match fs::read_dir("/scheme/acpi/tables") {
+ Ok(dir) => dir,
+ Err(e) => {
+ log::debug!("pcid: cannot read /scheme/acpi/tables: {e} (acpid running?)");
+ return Err(e);
+ }
+ };
- // TODO: validate/print MCFG?
+ let mut found_tables: Vec<String> = Vec::new();
for table_direntry in table_dir {
let table_path = table_direntry?.path();
- // Every directory entry has to have a filename unless
- // the filesystem (or in this case acpid) misbehaves.
- // If it misbehaves we have worse problems than pcid
- // crashing. `as_encoded_bytes()` returns some superset
- // of ASCII, so directly comparing it with an ASCII name
- // is fine.
let table_filename = table_path.file_name().unwrap().as_encoded_bytes();
+ found_tables.push(String::from_utf8_lossy(table_filename).into_owned());
+
if table_filename.get(0..4) == Some(&MCFG_NAME) {
let bytes = fs::read(table_path)?.into_boxed_slice();
match Mcfg::parse(&*bytes) {
@@ -174,6 +176,7 @@
return f(allocs, Vec::new(), [u32::MAX, u32::MAX, u32::MAX, u32::MAX]);
}
None => {
+ log::warn!("pcid: MCFG table found but failed to parse ({} bytes)", bytes.len());
return Err(io::Error::new(
io::ErrorKind::InvalidData,
"couldn't find mcfg table",
@@ -183,6 +186,12 @@
}
}
+ log::debug!(
+ "pcid: MCFG not found among {} ACPI table(s): {:?}",
+ found_tables.len(),
+ found_tables
+ );
+
Err(io::Error::new(
io::ErrorKind::NotFound,
"couldn't find mcfg table",
@@ -219,6 +228,24 @@
pub interrupt_map_mask: [u32; 4],
fallback: Pci,
}
+
+/// Validate an MCFG allocation entry address (cross-referenced with Linux
+/// `acpi_mcfg_valid_entry` in arch/x86/pci/mmconfig-shared.c).
+///
+/// - Addresses below 4 GiB are valid for legacy PCIe ECAM.
+/// - Addresses at or above 4 GiB require the MCFG table revision >= 1, matching
+/// the ACPI 4.0+ rule that 64-bit ECAM addresses are only valid when the
+/// firmware signals support via the revision field.
+fn validate_mcfg_addr(addr: u64) -> Result<(), String> {
+ const FOUR_GIB: u64 = 0x1_0000_0000;
+
+ if addr < FOUR_GIB {
+ return Ok(());
+ }
+
+ Err(format!("address {addr:#018x} >= 4 GiB requires MCFG revision >= 1 (ACPI 4.0+); entry skipped"))
+}
+
struct Alloc {
seg: u16,
start_bus: u8,
@@ -239,9 +266,11 @@
Ok(pcie) => pcie,
Err(fdt_error) => {
log::warn!(
- "Couldn't retrieve PCIe info, perhaps the kernel is not compiled with \
- acpi or device tree support? Using the PCI 3.0 configuration space \
- instead. ACPI error: {:?} FDT error: {:?}",
+ "PCIe (ECAM/MCFG) not available: {}. \
+ Device tree ECAM also not found: {}. \
+ Falling back to PCI 3.0 config space (I/O ports 0xCF8/0xCFC). \
+ For PCI Express support, use QEMU Q35 machine type (-machine q35) \
+ or ensure your platform firmware provides an MCFG ACPI table.",
acpi_error,
fdt_error
);
@@ -266,24 +295,44 @@
.0
.iter()
.filter_map(|desc| {
+ if desc.seg_group_num != 0 {
+ let seg = desc.seg_group_num;
+ let start = desc.start_bus;
+ let end = desc.end_bus;
+ log::warn!(
+ "pcid: skipping MCFG entry at seg={seg} bus {start}..={end}: multi-segment not yet implemented",
+ );
+ return None;
+ }
+
+ if let Err(reason) = validate_mcfg_addr(desc.base_addr) {
+ let addr = desc.base_addr;
+ let seg = desc.seg_group_num;
+ let start = desc.start_bus;
+ let end = desc.end_bus;
+ log::warn!(
+ "pcid: skipping MCFG entry at {addr:#018x} (seg={seg} bus {start}..={end}): {reason}",
+ );
+ return None;
+ }
+
+ let seg = desc.seg_group_num;
+ let start = desc.start_bus;
+ let end = desc.end_bus;
Some(Alloc {
- seg: desc.seg_group_num,
- start_bus: desc.start_bus,
- end_bus: desc.end_bus,
+ seg,
+ start_bus: start,
+ end_bus: end,
mem: PhysBorrowed::map(
desc.base_addr.try_into().ok()?,
BYTES_PER_BUS
- * (usize::from(desc.end_bus) - usize::from(desc.start_bus) + 1),
+ * (usize::from(end) - usize::from(start) + 1),
Prot::RW,
MemoryType::Uncacheable,
)
.inspect_err(|err| {
log::error!(
- "failed to map seg {} bus {}..={}: {}",
- { desc.seg_group_num },
- { desc.start_bus },
- { desc.end_bus },
- err
+ "failed to map seg {seg} bus {start}..={end}: {err}",
)
})
.ok()?,
@@ -0,0 +1,15 @@
diff --git a/bootstrap/src/procmgr.rs b/bootstrap/src/procmgr.rs
index ce37d57..b1268e0 100644
--- a/bootstrap/src/procmgr.rs
+++ b/bootstrap/src/procmgr.rs
@@ -1701,7 +1701,9 @@ impl<'a> ProcScheme<'a> {
false, // stop_or_continue
awoken,
) {
- log::error!("failed to send SIGCHLD to parent PID {ppid:?}: {err}");
+ // EPERM on SIGCHLD to PID 1 is a known kernel limitation.
+ // The procmgr continues correctly after this; downgrade to debug.
+ log::debug!("failed to send SIGCHLD to parent PID {ppid:?}: {err}");
}
if let Some(init_rc) = self.processes.get(&INIT_PID) {
@@ -0,0 +1,42 @@
--- a/drivers/input/ps2d/src/mouse.rs
+++ b/drivers/input/ps2d/src/mouse.rs
@@ -1,8 +1,8 @@
use crate::controller::Ps2;
use std::time::Duration;
-pub const RESET_RETRIES: usize = 10;
-pub const RESET_TIMEOUT: Duration = Duration::from_millis(1000);
+pub const RESET_RETRIES: usize = 3;
+pub const RESET_TIMEOUT: Duration = Duration::from_millis(250);
pub const COMMAND_TIMEOUT: Duration = Duration::from_millis(100);
#[derive(Clone, Copy, Debug)]
@@ -260,7 +264,8 @@
MouseResult::Timeout(COMMAND_TIMEOUT)
} else {
log::warn!("unknown mouse response {:02X} after reset", data);
- self.reset(ps2)
+ *self = MouseState::None;
+ MouseResult::None
}
}
MouseState::Bat => {
@@ -352,12 +357,14 @@
self.reset(ps2)
}
MouseState::Reset => {
- log::warn!("timeout waiting for mouse reset");
- self.reset(ps2)
+ log::debug!("timeout waiting for mouse reset, fast-failing");
+ *self = MouseState::None;
+ MouseResult::None
}
MouseState::Bat => {
- log::warn!("timeout waiting for BAT completion");
- self.reset(ps2)
+ log::debug!("timeout waiting for BAT completion, fast-failing");
+ *self = MouseState::None;
+ MouseResult::None
}
MouseState::IdentifyTouchpad { .. } => {
//TODO: retry?
@@ -25,20 +25,20 @@
+ if num { led_byte |= 2; }
+ if caps { led_byte |= 4; }
+ if let Err(err) = self.keyboard_command_inner(0xED) {
+ warn!("ps2d: failed to send LED command 0xED: {:?}", err);
+ log::debug!("ps2d: LED command 0xED not supported: {:?}", err);
+ return;
+ }
+ match self.read_timeout(DEFAULT_TIMEOUT) {
+ Ok(0xFA) => {
+ if let Err(err) = self.write(led_byte) {
+ warn!("ps2d: failed to send LED byte {:02X}: {:?}", led_byte, err);
+ log::debug!("ps2d: failed to send LED byte {:02X}: {:?}", led_byte, err);
+ }
+ }
+ Ok(val) => {
+ warn!("ps2d: LED command ACK expected 0xFA, got {:02X}", val);
+ log::debug!("ps2d: LED command ACK expected 0xFA, got {:02X}", val);
+ }
+ Err(err) => {
+ warn!("ps2d: LED command ACK timeout: {:?}", err);
+ log::debug!("ps2d: LED command ACK timeout: {:?}", err);
+ }
+ }
+ }