# BST-TIMEOUT: 600   # five awk passes over 9.1 MB of tables, the largest 5.5 MB;
#                    # measured 61s on target (==BST-SUITE, 2026-08-20), so 600
#                    # is ~10x headroom -- sized for the host's worst day
. "$BST_LIB"

# hwdata is data only -- the numeric-ID-to-name tables that every
# device-inspection tool on a Linux system reads. There is no tool to `require`:
# what ships is five tables plus a modprobe blacklist and a pkg-config file, and
# not one byte of the package is code.
#
# THIS SUITE PARSES, IT DOES NOT STAT. Asserting that five files exist is
# compatible with every one of them being a half-fetched prefix, which is
# precisely the failure an allarch data recipe produces when it produces one.
# So each table is read with its OWN grammar, its records are counted, its
# version stamp is checked against the SRCREV that froze it, and a handful of
# real IDs are resolved end to end -- with a negative control beside each lookup,
# because a filter that had silently stopped filtering would answer the first
# record to every query and "Intel Corporation" would then prove nothing.
#
# The counts are EXACT rather than floors. They are frozen by the recipe's
# SRCREV (44635308, hwdata 0.381), the rootfs is read-only at browser runtime so
# nothing here can repair a short file, and an exact count is the only form that
# catches a table truncated in its MIDDLE -- a floor and a `test -s` both pass on
# one of those. A recipe bump is expected to move them, deliberately and
# visibly; that is the point of pinning them.

HW=/usr/share/hwdata
PC=/usr/share/pkgconfig/hwdata.pc
MPD=/usr/lib/modprobe.d/dist-blacklist.conf

# --- section 1: the payload arrived, whole ----------------------------------
# `wc -c "$f"` and not `wc -c < "$f"`: a top-level stdin redirect from the
# virtiofs root makes dup2(2) return the kernel's restart-park sentinel as 513,
# and ash treats that as fatal and EXITS -- taking every case below this line
# with it while the run still reads as a clean pass (brintos/linux#73, and the
# packager README's table of the two shapes).
size() { LC_ALL=C wc -c "$1" 2>/dev/null | LC_ALL=C awk '{print $1; exit}'; }

check hwdata/datadir test -d "$HW"

check_eq hwdata/size-pci-ids   1444058 "$(size "$HW/pci.ids")"
check_eq hwdata/size-usb-ids    728224 "$(size "$HW/usb.ids")"
check_eq hwdata/size-oui-txt   5510620 "$(size "$HW/oui.txt")"
check_eq hwdata/size-iab-txt   1779268 "$(size "$HW/iab.txt")"
check_eq hwdata/size-pnp-ids     63075 "$(size "$HW/pnp.ids")"

# The one file the package installs outside ${datadir}. It is why the recipe
# passes --libdir at all, and it is the half of FILES:${PN} that a datadir-only
# packaging change would silently drop.
check    hwdata/modprobe-d      test -f "$MPD"
check_eq hwdata/size-blacklist     884 "$(size "$MPD")"

# Nothing else ships a /usr/share/hwdata, so the directory holds exactly the
# five tables and nothing has been packaged into it by accident.
check_eq hwdata/datadir-count 5 \
  "$(LC_ALL=C ls -1 "$HW" 2>/dev/null | LC_ALL=C awk 'END{print NR+0}')"

# Everything below needs a parser. If awk is somehow absent the lookups SKIP
# with the reason rather than reporting a data fault they never measured.
if ! command -v awk >/dev/null 2>&1; then
  bst_skip hwdata/parse "no awk on this image; nothing here can read the tables"
  exit 0
fi

# Pull one tagged key out of a captured tagged block. The tables are read ONCE
# each into these blocks -- oui.txt alone is 5.5 MB and a pass per assertion
# would be twenty passes over it.
val() { printf '%s\n' "$2" | LC_ALL=C awk -v k="$1" \
          'index($0,k"=")==1{print substr($0,length(k)+2); exit}'; }

# --- section 2: pci.ids ------------------------------------------------------
# Grammar: `xxxx  Vendor` at column 0, `\txxxx  Device` under it, `\t\txxxx yyyy
#   Subsystem` under that, and a separate class block introduced by `C xx  Name`.
# The parse is STATE-AWARE: device lines are counted only while inside a vendor
# section, so the class block's own two-hex subclass lines cannot be mistaken for
# devices. `bad` counts every non-comment, non-blank line the grammar does not
# account for -- it is the assertion that this is a complete parse of the file
# and not a lucky pattern match over part of it.
PCI=$(LC_ALL=C awk '
/^#/ { next }
/^[ \t]*$/ { next }
substr($0,1,1)=="\t" {
  if (substr($0,2,1)=="\t") {
    if (inv && match(substr($0,3), /^[0-9a-f][0-9a-f][0-9a-f][0-9a-f] [0-9a-f][0-9a-f][0-9a-f][0-9a-f]  /)) sb++
    next
  }
  if (inv && match(substr($0,2), /^[0-9a-f][0-9a-f][0-9a-f][0-9a-f]  /)) {
    dev++
    id = substr($0,2,4)
    if (curv=="8086" && id=="1237") d1 = substr($0,8)
    if (curv=="1af4" && id=="1000") d2 = substr($0,8)
    if (curv=="8086" && id=="ffff") dn = substr($0,8)
  }
  next
}
match($0, /^[0-9a-f][0-9a-f][0-9a-f][0-9a-f]  /) {
  ven++; inv=1; curv = substr($0,1,4)
  if (curv=="8086") v1 = substr($0,7)
  if (curv=="10de") v2 = substr($0,7)
  if (curv=="zzzz") vn = substr($0,7)
  next
}
match($0, /^C [0-9a-f][0-9a-f]  /) { cls++; inv=0; next }
{ bad++ }
/^#[ \t]*Version:/ { }
END {
  printf "ven=%d\ndev=%d\nsub=%d\ncls=%d\nbad=%d\n", ven+0, dev+0, sb+0, cls+0, bad+0
  printf "v8086=%s\nv10de=%s\nvabsent=%s\n", v1, v2, vn
  printf "d8086_1237=%s\nd1af4_1000=%s\ndabsent=%s\n", d1, d2, dn
}' "$HW/pci.ids" 2>&1)

check_eq pci-ids/vendors      2378  "$(val ven "$PCI")"
check_eq pci-ids/devices     18685  "$(val dev "$PCI")"
check_eq pci-ids/subsystems  16183  "$(val sub "$PCI")"
check_eq pci-ids/classes        22  "$(val cls "$PCI")"
# Zero unaccounted lines: every byte of the file fell into one of the four
# productions above. This is what makes the counts a measurement of the table
# rather than of the regex.
check_eq pci-ids/grammar-complete 0 "$(val bad "$PCI")"

# The lookups the ticket asks for -- a vendor, and one of that vendor's devices.
check_eq pci-ids/vendor-8086 "Intel Corporation"            "$(val v8086 "$PCI")"
check_eq pci-ids/vendor-10de "NVIDIA Corporation"           "$(val v10de "$PCI")"
check_eq pci-ids/device-8086-1237 "440FX - 82441FX PMC [Natoma]" "$(val d8086_1237 "$PCI")"
# A device under a DIFFERENT vendor, so the device lookup is shown to be scoped
# by the vendor section it sits in and not to be a file-wide search for "1000".
check_eq pci-ids/device-1af4-1000 "Virtio network device"   "$(val d1af4_1000 "$PCI")"
# The negative controls for both levels.
check_eq pci-ids/vendor-absent ""  "$(val vabsent "$PCI")"
check_eq pci-ids/device-absent ""  "$(val dabsent "$PCI")"

# The version stamp is the PCI ID Project's own, frozen by the SRCREV. It is NOT
# the recipe's PV -- see section 6.
check_out pci-ids/version  "Version: 2024.03.31" head -6 "$HW/pci.ids"
check_out pci-ids/date     "Date:    2024-03-31" head -6 "$HW/pci.ids"

# --- section 3: usb.ids ------------------------------------------------------
# Same two-level vendor/device grammar, but the file continues past the vendor
# table into ten prefixed section blocks (C, AT, HID, R, BIAS, PHY, HUT, L, VT,
# HCC) whose entries are also hex-and-two-spaces. `inv` is cleared by any
# unindented line that is not a vendor, which is what keeps a HID usage or a
# language code out of the device count.
USB=$(LC_ALL=C awk '
/^#/ { next }
/^[ \t]*$/ { next }
substr($0,1,1)=="\t" {
  if (substr($0,2,1)=="\t") { next }
  if (inv && match(substr($0,2), /^[0-9a-f][0-9a-f][0-9a-f][0-9a-f]  /)) {
    dev++
    id = substr($0,2,4)
    if (curv=="1d6b" && id=="0002") d1 = substr($0,8)
    if (curv=="046d" && id=="c31c") d2 = substr($0,8)
    if (curv=="1d6b" && id=="ffff") dn = substr($0,8)
  }
  next
}
match($0, /^[0-9a-f][0-9a-f][0-9a-f][0-9a-f]  /) {
  ven++; inv=1; curv = substr($0,1,4)
  if (curv=="1d6b") v1 = substr($0,7)
  if (curv=="046d") v2 = substr($0,7)
  next
}
{ inv=0; sect++ }
END {
  printf "ven=%d\ndev=%d\nsect=%d\n", ven+0, dev+0, sect+0
  printf "v1d6b=%s\nv046d=%s\n", v1, v2
  printf "d1d6b_0002=%s\nd046d_c31c=%s\ndabsent=%s\n", d1, d2, dn
}' "$HW/usb.ids" 2>&1)

check_eq usb-ids/vendors 3423  "$(val ven "$USB")"
check_eq usb-ids/devices 20464 "$(val dev "$USB")"
# The ten section headers the file carries after the vendor table. Pinned so a
# truncation that lopped the tail off would not read as a clean vendor table.
check_eq usb-ids/sections 306  "$(val sect "$USB")"

check_eq usb-ids/vendor-1d6b "Linux Foundation" "$(val v1d6b "$USB")"
check_eq usb-ids/vendor-046d "Logitech, Inc."   "$(val v046d "$USB")"
check_eq usb-ids/device-1d6b-0002 "2.0 root hub" "$(val d1d6b_0002 "$USB")"
check_eq usb-ids/device-046d-c31c "Keyboard K120" "$(val d046d_c31c "$USB")"
check_eq usb-ids/device-absent "" "$(val dabsent "$USB")"

check_out usb-ids/version "Version: 2024.03.18" head -14 "$HW/usb.ids"

# THE OTHER usb.ids ON THIS IMAGE. usbids (meta-oe recipes-support/usbids, the
# same porting round) installs a bare /usr/share/usb.ids -- a DIFFERENT and newer
# table at a different path. The two packages' file lists are disjoint, which is
# the whole of why they can coexist, and this pair of cases is where that is
# asserted rather than assumed. If usbids is not on this image the second case
# skips; what must never happen is the two ending up at one path.
check hwdata/usb-ids-own-path test -f "$HW/usb.ids"
if [ -f /usr/share/usb.ids ]; then
  a=$(size "$HW/usb.ids"); b=$(size /usr/share/usb.ids)
  if [ "$a" != "$b" ]; then bst_ok hwdata/usb-ids-distinct-from-usbids
  else bst_notok hwdata/usb-ids-distinct-from-usbids \
    "/usr/share/usb.ids and $HW/usb.ids are the same size ($a) -- one may have overwritten the other"; fi
else
  bst_skip hwdata/usb-ids-distinct-from-usbids \
    "usbids is not on this image; hwdata owns the only usb.ids"
fi

# --- section 4: oui.txt and iab.txt -----------------------------------------
# The IEEE registries. Grammar: a record is a `XX-XX-XX   (hex)   Organization`
# line followed by a `XXXXXX   (base 16)   Organization` line and an indented
# address block. Counting BOTH forms and asserting they agree is the
# cross-representation check -- a file cut mid-record would leave the two counts
# one apart, which neither count alone would show.
OUI=$(LC_ALL=C awk '
match($0, /^[0-9A-F][0-9A-F]-[0-9A-F][0-9A-F]-[0-9A-F][0-9A-F][ \t]+\(hex\)/) {
  n++; pre = substr($0,1,8); org = $0; sub(/^.*\(hex\)[ \t]+/, "", org)
  if (pre=="00-50-56") a = org
  if (pre=="00-1B-21") b = org
  if (pre=="08-00-27") c = org
  if (pre=="ZZ-ZZ-ZZ") z = org
  next
}
match($0, /^[0-9A-F][0-9A-F][0-9A-F][0-9A-F][0-9A-F][0-9A-F][ \t]+\(base 16\)/) { m++; next }
END { printf "hex=%d\nbase16=%d\nvmware=%s\nintel=%s\npcs=%s\nabsent=%s\n", n+0, m+0, a, b, c, z }
' "$HW/oui.txt" 2>&1)

check_eq oui/records 35297 "$(val hex "$OUI")"
check_eq oui/hex-base16-agree "$(val hex "$OUI")" "$(val base16 "$OUI")"
# The MAC-prefix lookup the ticket asks for, three real assignments and a control.
check_eq oui/00-50-56 "VMware, Inc."           "$(val vmware "$OUI")"
check_eq oui/00-1B-21 "Intel Corporate"        "$(val intel "$OUI")"
check_eq oui/08-00-27 "PCS Systemtechnik GmbH" "$(val pcs "$OUI")"
check_eq oui/absent   ""                       "$(val absent "$OUI")"
check_out oui/header "OUI/MA-L" head -2 "$HW/oui.txt"

# iab.txt is the MA-S registry: a 24-bit prefix SHARED between assignees, each
# holding a 12-bit range of it. Its second line is therefore a RANGE
# (`0D7000-0D7FFF`), not a bare base-16 prefix, which is the one grammatical
# difference from oui.txt and the reason it gets its own parse.
IAB=$(LC_ALL=C awk '
match($0, /^[0-9A-F][0-9A-F]-[0-9A-F][0-9A-F]-[0-9A-F][0-9A-F][ \t]+\(hex\)/) {
  n++; org = $0; sub(/^.*\(hex\)[ \t]+/, "", org)
  if (n==1) { p1 = substr($0,1,8); o1 = org }
  next
}
match($0, /^[0-9A-F][0-9A-F][0-9A-F][0-9A-F][0-9A-F][0-9A-F]-[0-9A-F][0-9A-F][0-9A-F][0-9A-F][0-9A-F][0-9A-F][ \t]+\(base 16\)/) {
  m++; if (m==1) { r1 = $0; sub(/[ \t]+\(base 16\).*$/, "", r1) }
  next
}
END { printf "hex=%d\nrange=%d\nfirstpre=%s\nfirstorg=%s\nfirstrange=%s\n", n+0, m+0, p1, o1, r1 }
' "$HW/iab.txt" 2>&1)

check_eq iab/records 4575 "$(val hex "$IAB")"
check_eq iab/hex-range-agree "$(val hex "$IAB")" "$(val range "$IAB")"
check_eq iab/first-prefix "40-D8-55"            "$(val firstpre "$IAB")"
check_eq iab/first-org    "Avant Technologies"  "$(val firstorg "$IAB")"
check_eq iab/first-range  "0D7000-0D7FFF"       "$(val firstrange "$IAB")"
check_out iab/header "IAB Range" head -2 "$HW/iab.txt"

# --- section 5: pnp.ids -----------------------------------------------------
# The legacy EISA/PnP three-letter manufacturer codes -- a flat
# `AAA<TAB>Organization` table with no header and no version stamp of any kind,
# which is why it is counted and looked up but has no version case.
PNP=$(LC_ALL=C awk '
match($0, /^[A-Z][A-Z][A-Z]\t/) {
  n++; id = substr($0,1,3)
  if (id=="IBM") a = substr($0,5)
  if (id=="DEL") b = substr($0,5)
  next
}
/^[ \t]*$/ { next }
{ bad++; if (bad<=2) badline[bad] = substr($0,1,3) }
END { printf "n=%d\nbad=%d\nibm=%s\ndel=%s\nbad1=%s\nbad2=%s\n", n+0, bad+0, a, b, badline[1], badline[2] }
' "$HW/pnp.ids" 2>&1)

check_eq pnp/records 2544 "$(val n "$PNP")"
check_eq pnp/IBM "IBM Brasil" "$(val ibm "$PNP")"
check_eq pnp/DEL "Dell Inc."  "$(val del "$PNP")"

# TWO RECORDS IN pnp.ids DO NOT MATCH ITS OWN GRAMMAR, and that is upstream's
# table rather than a fault in this port: `inu` is lower-case and `VQ@` is not
# three letters at all. Both are in the source at this SRCREV, both reproduce on
# any host, and neither is anything brintOS can fix -- so they are PINNED here,
# by count and by id, instead of being left to read as a red or quietly excluded
# by a laxer regex. The day upstream corrects them this case goes red, which is
# the prompt to re-read this comment and drop the exception.
check_eq pnp/malformed-upstream-count 2 "$(val bad "$PNP")"
check_eq pnp/malformed-upstream-1 "inu" "$(val bad1 "$PNP")"
check_eq pnp/malformed-upstream-2 "VQ@" "$(val bad2 "$PNP")"

# --- section 6: the pkg-config file -----------------------------------------
# THIS IS THE ONE WAY AN allarch DATA RECIPE GENUINELY GOES WRONG. hwdata.pc is
# generated on the BUILD host at do_configure from hwdata.pc.in; it is the single
# file in the package whose contents could carry a build-host path instead of a
# target one, and nothing about the tables above would show it if they did.
check hwdata/pc-file test -f "$PC"

if ! command -v pkg-config >/dev/null 2>&1; then
  bst_skip hwdata/pc-query "no pkg-config on this image; the .pc cannot be resolved here"
else
  # `--modversion` reads the Version field, which comes from hwdata's own
  # release number and IS the recipe's PV. This is the assertion against PV;
  # the table stamps in sections 2-3 are assertions against the SRCREV, and
  # comparing either to the other would be asserting a relationship that does
  # not exist -- hwdata is a redistribution and its tables are older than it.
  check_eq hwdata/pc-modversion "0.381" "$(LC_ALL=C pkg-config --modversion hwdata 2>&1)"

  pkgdatadir=$(LC_ALL=C pkg-config --variable=pkgdatadir hwdata 2>&1)

  # THE LITERAL ANSWER CARRIES A DOUBLED LEADING SLASH, and it is pinned rather
  # than normalised away. hwdata.pc spells the variable
  # `pkgdatadir=${pc_sysrootdir}${datadir}/hwdata` so that pkgconf substitutes a
  # sysroot at QUERY time instead of baking one in at configure time, and pkgconf
  # defaults pc_sysrootdir to "/" when the environment names none -- so the
  # answer is `//usr/share/hwdata`. Linux resolves that identically to the
  # single-slash form. Pinning it means a change in either half of that
  # mechanism is visible here instead of being absorbed by a tolerant matcher.
  check_eq hwdata/pc-pkgdatadir-literal "//usr/share/hwdata" "$pkgdatadir"

  # And the assertion that actually matters: whatever string it answered, the
  # directory that string names is the one the payload landed in. This is the
  # case that goes red if a build-host path is ever baked into the .pc -- the
  # literal case above would go red too, but this one says WHY it matters.
  check hwdata/pc-pkgdatadir-resolves test -f "$pkgdatadir/pci.ids"
  check_eq hwdata/pc-pkgdatadir-is-datadir "$(size "$HW/pci.ids")" "$(size "$pkgdatadir/pci.ids")"

  # No build-host path anywhere in the shipped file. The tmp-<arm> work tree and
  # the recipe workdir are the two shapes a leaked configure-time path takes.
  leak=$(LC_ALL=C awk '/tmp-glibc|tmp-musl|\/work\/|recipe-sysroot|\/build-/{n++} END{print n+0}' "$PC" 2>&1)
  check_eq hwdata/pc-no-buildhost-path 0 "$leak"

  # hwdata.pc is reachable on pkgconf's DEFAULT search path with no
  # PKG_CONFIG_PATH set, which is not a given: it installs to
  # ${datadir}/pkgconfig, not ${libdir}/pkgconfig where every compiled port on
  # this image puts its .pc, and poky's pkgconf is configured with both only
  # because of --with-pkg-config-dir. Asserted by NAME resolution, which is the
  # part that belongs to this port and passes.
  check hwdata/pc-default-search-path env -u PKG_CONFIG_PATH -u PKG_CONFIG_LIBDIR \
    pkg-config --exists hwdata

  # THIS CASE IS EXPECTED RED, AND IT IS NOT ABOUT hwdata.
  #
  # Enumerating the default search path is the stronger statement -- it says
  # hwdata is visible to a consumer that DISCOVERS packages rather than one that
  # already knows the name -- and it cannot be made on this image today, because
  # `pkg-config --list-all` dies partway through with SIGILL (rc 132) after
  # printing ~337 of the image's 505 packages. The cause is nothing to do with
  # this port: the coroutine frame arena bump-extends to a 252 MB high-water on
  # ELEVEN live frames over ~1.04M allocations, hits the 256 MiB per-process
  # userMaxBytes ceiling, and traps instead of returning failure the way an
  # ordinary malloc at the very same ceiling does (gawk reports a clean ENOMEM
  # in the same boot). Measured with hwdata excluded entirely
  # (PKG_CONFIG_LIBDIR=/usr/lib/pkgconfig, 499 packages, none of them hwdata's)
  # -- it reds identically, so hwdata neither causes nor worsens it.
  #
  # BRINTOS-WORKAROUND: brintos/linux#95 remove -- left RED and attributed
  # rather than skipped or asserted-as-broken, so it flips GREEN by itself the
  # day the frame arena stops trapping on OOM. When it does, delete this comment
  # and this note in docs/brintos-test-plan.md 364.1; do not "fix" the case by
  # weakening it.
  check_out hwdata/pc-list-all "hwdata" pkg-config --list-all
fi

# --- section 7: what this image deliberately does NOT carry ------------------
# Recorded rather than left to be discovered, the same way 99-iso-codes.test
# records its missing locale catalogues. The natural readers of these tables are
# lspci and lsusb; neither pciutils nor usbutils is ported, and a target with no
# PCI or USB bus to enumerate would have nothing for them to report anyway. So
# the image gains a data set and no new behaviour in anything already installed.
# A skip, not a red -- and the place that decision is written down.
if command -v lspci >/dev/null 2>&1 || command -v lsusb >/dev/null 2>&1; then
  bst_ok hwdata/consumer-present
else
  bst_skip hwdata/consumer-present \
    "no lspci/lsusb on this image (pciutils and usbutils are not ported); the tables ship for future consumers"
fi