startergo/pcitree.sh

## pcitree.sh
#!/bin/bash
# by joevt Jan 22, 2023

#===================
if [[ $EUID -ne 0 ]]; then
   echo "This script must be run as root like this:"
   echo "sudo $0"
   exit 1
fi
#===================
style="wide"
style="narrow"

#encoding="ascii"
encoding="utf8"

#method="-H1" # -A intel-conf1
method=""    # default method which for macOS is -A darwin

#===================
if [ $encoding == "ascii" ]; then
	box0='\'
	box1='-'
	box2='\'
	box3='+'
	box4='|'
	box5='.'
else
	box0='┐'
	box1='─'
	box2='└'
	box3='├'
	box4='│'
	box5='┬'
fi

if [ $style == "wide" ]; then

	if [ $encoding == "ascii" ]; then
		# since we're using \, we want to indent one extra space
		rootindent='            '
		subindent='                '
	else
		rootindent='           '
		subindent='               '
	fi

	domainprefix="\\${box1}\\["
	domainsuffix="\\]\\${box1}\\${box0}"

	bridgeprefix="\\${box3}\\${box1}"
	bridgesuffixlong="\\${box1}\\${box1}\\${box1}\\${box1}\\${box0}"
	bridgesuffix="\\${box1}\\${box0}"

	deviceprefix="\\${box3}\\${box1}"
	devicesuffix=""
else
	rootindent=''
	subindent=' '

	domainprefix="\\${box5}\\["
	domainsuffix="\\]"

	bridgeprefix="\\${box3}\\${box5}"
	bridgesuffixlong=""
	bridgesuffix=""

	deviceprefix="\\${box3}\\${box1}"
	devicesuffix=""

	fulladdress=1
fi

#===================

# Get header type and bus information for all pci devices - reverse sort so last bus is first (required for device path substitution step below).
# Format is "domain:bus%device.function domain:primary domain:secondary domain:subordinate".
# Keep the domain number for device path substitution step below.
# -0777 allows perl to process all lines at once - here we process two lines at a time.
setpci $method -v -s '*:*:*' HEADER_TYPE.b 0x18.l 2> /dev/null | perl -0777 -pe '
	s/((.{4}):(..):(..\..)) \@0e = ([08][12])\n\1 \@18 = ..(.{2})(.{2})(.{2})\n/$2:$3%$4 $2:$8 $2:$7 $2:$6\n/g; # bridge(01) or card bus(02) header type, multifunction(80) or not (00).
	s/((.{4}):(..):(..\..)) \@0e = (..)\n\1 \@18 = ..(.{2})(.{2})(.{2})\n/$2:$3%$4\n/g # endpoint (or anything else) header type
' \
| sort -r > /tmp/pcilist.txt

# If primary = secondary = subordinate then that is a problem - change them to XX.
sed '/^\(....:..%..\..\) \(\(.\{4\}:\)..\) \2 \2$/s//\1 \3XX \3XX \3XX/' /tmp/pcilist.txt > /tmp/pcilist2.txt

echo '/ ....:/s// /g' > /tmp/pcipaths.txt # Add a command that will remove domain from primary, secondary, and subordinate bus numbers.

# Create device path substitution commands that will add parent bus to each secondary bus.
# When applied to the list of pci devices, the commands will create a full device path for
# each pci device. Multiple commands will be applied to each device to build the device
# path from right to left: end point (greatest bus number) to domain (lowest bus number).
# The commands are sorted from highest bus number to lowest bus number to make that possible.
sed -nE '/^(....:..%..\..) .{4}:.. (.{4}:..) .*/s//\/(\2)\/s\/\/\1%\\1\/g/p' /tmp/pcilist2.txt >> /tmp/pcipaths.txt
sed -E -f /tmp/pcipaths.txt /tmp/pcilist2.txt > /tmp/pcitree1.txt # Apply the device path substitution commands.
sed -E '/^(....:..).*/s//\1/g' /tmp/pcitree1.txt | sort -u > /tmp/pcitree2.txt # Get all root complexes (usually just 0000:00).

cat /tmp/pcitree1.txt /tmp/pcitree2.txt | LC_ALL=C sort > /tmp/pcitree3.txt # Concatenate the devices and root complexes and sort.

# Use perl to do some formatting.
perl -CSDA -Mutf8 -pe '
	s/^(.*(\w\w\w\w:\w\w%\w\w\.\w).*)/\1  # \2/;                                             # save domain/bus/device/function after a "  # " - this will be used to append name and ids
	s/^\w\w\w\w:\w\w%/'"$rootindent"'/;                                                      # remove domain/bus number from left side and indent
	s/\w\w\.\w%\w\w\w\w:\w\w%/'"$subindent"'/g;                                              # use the paths to do more indenting
	s/(\w\w\.\w) \w\w (\w\w) \2/'"${bridgeprefix}"'\1-[\2]'"${bridgesuffixlong}"'/;          # format the secondary and subordinate bus numbers for single bus bridge
	s/(\w\w\.\w) \w\w (\w\w) (\w\w)/'"${bridgeprefix}"'\1-[\2-\3]'"${bridgesuffix}"'/;       # format the secondary and subordinate bus numbers for multiple bus bridge
	s/%/:/;                                                                                  # replace % bus/device delimiter with standard ":"
	s/(\w\w\.\w  )/'"${deviceprefix}"'\1'"${devicesuffix}"'/;                                # add the "+-" formating for devices
	s/^(\w\w\w\w:\w\w)$/'"${domainprefix}"'\1'"${domainsuffix}"'/                            # add the -[ ]\" for domain/bus
' \
/tmp/pcitree3.txt > /tmp/pcitree4.txt

# https://perldoc.perl.org/perlrun.html
# The -C flag controls some of the Perl Unicode features.
#      S: STDIN, STDOUT and STDERR are assumed to be in UTF-8.
#      D: UTF-8 is the default PerlIO layer for both input and output streams.
#      A: @ARGV elements are expected to be strings encoded in UTF-8.
# -0[octal/hexadecimal] specifies the input record separator ($/ ) as an octal or hexadecimal number.
#      Any value 0400 or above will cause Perl to slurp files whole, but by convention the value 0777 is the one normally used for this purpose.
# -Mmodule executes use module ; before executing your program. This loads the module and calls its import method, causing the module to have its default effect, typically importing subroutines or giving effect to a pragma.
# -p causes Perl to assume a loop around your program, which makes it iterate over filename arguments somewhat like sed.
# -e commandline May be used to enter one line of program.
perl -CSDA -Mutf8 -0777 -pe '

	#	add | between devices of the same bus
	do {
		$good=0;
		while (
			/											#	begin regular expression
				(										# 		begin g1
					\n									#			\n
					([\ '"\\${box4}"']*)				#			[ |]* -> g2
					['"\\${box3}\\${box4}"']			#			[+|]
					[^\n]*								#			rest of the line up to \n, (change + to *+ to give nothing back)
					(?{$X=pos()})						#			mark position
					\n									#			\n
				)										#		end g1
				(?=										#		begin lookahead
					(									#			begin group
						\2\ [^\n]*						#				following line(s) start with g2 also (change + to *+ to give nothing back)
						\n								#
					)+									#			end group, match one or more, (change + to ++ to give nothing back)
					\2									#			the line after those start with g2
					'"\\${box3}"'						#				 and "+" meaning that another device follows g1 on the same bus
				)										#		end lookahead
				(
					\2									#
					(?{$C=pos()})						#		$C is the position where we want to insert the |
					\ 									#		the character at position $C to be modified is a space character
				)										#
			/gx											#	end regular expression (g=global; x=ignore white space in RE)
		) {
			$good = 1;
			substr($_,$C,1) = "'"\\${box4}"'";			#	replace the space character at position $C with a |
			pos() = $X;									#	start next search at $X which is the end of the previous line
		}
	} until ($good == 0);

	# replace + with \ for last device of a bus
	s/(\n[\ '"\\${box4}"']*)'"\\${box3}"'(?![^\n]*\1['"\\${box3}\\${box4}"'])/\1'"\\${box2}"'/g;
' /tmp/pcitree4.txt > /tmp/pcitree5.txt

# Get the length of the longest line so we can align device information in a detail column.
spaces=$(sed -E 's/./ /g' /tmp/pcitree5.txt | sort -r | head -n 1)


lspci $method -nnv -D 2> /dev/null > /tmp/pcidevices0_lspci.txt
perl -pe '
	s/^\n//g;
	s/^[ \t].*\n//g;
	s|/|\\/|g;
	s|^(....:..:..\..) ([^[]*) (\[....\])(.*) (\[....:....\])( \(rev ..\))?(?: \(prog-if ..( \[.*?\])\))?|/# (\1)\$/s//'"                   "'#                \\1 \5 \3:\6          : \2\7                                      \4/|;
	s/\]:([^:]{9}.*?) *:([^:]{37}.*?) *:/]\1\2 :/ # rev column is 9 characters and type column is 36 characters
' \
/tmp/pcidevices0_lspci.txt > /tmp/pcidevices1_lspci_sed.txt
echo '/^(.{'"$((${#spaces} - 14))"'}) *(.*)/s//\1\2/' >> /tmp/pcidevices1_lspci_sed.txt # create the substitutions for the detail column


# Get link width and speed max/current.
pat="(....:..:..\..) @09 = (..)"
IFS=$'\n'
echo "" > /tmp/pcidevices2_link_sed.txt
echo "" > /tmp/pcidevices1_progintf_sed.txt
for thedevice in $(sudo setpci $method -v -s '*:*:*' CLASS_PROG.b 2> /dev/null); do
	pcidevice="${thedevice% @*}"
	proginterface="${thedevice#* = }"
	# We have to do this in a loop because setpci stops if express capability doesn't exist
	LinkRegisters=$(sudo setpci $method -s $pcidevice CAP_EXP+12.w CAP_EXP+c.l 2> /dev/null)
	if [[ -n $LinkRegisters ]]; then
		LinkStatusRegister=${LinkRegisters:0:4}
		LinkWidth=$((0x$LinkStatusRegister >> 4 & 31))
		if [[ $LinkWidth -ge 0 ]]; then
			LinkSpeed=$((0x$LinkStatusRegister & 15))
			LinkCapabilitiesRegister=${LinkRegisters:5:8}
			MaxLinkWidth=$((0x$LinkCapabilitiesRegister >> 4 & 31))
			MaxLinkSpeed=$((0x$LinkCapabilitiesRegister & 15))
			echo "/#                ($pcidevice .*$)/s//# g${MaxLinkSpeed}x${MaxLinkWidth} > g${LinkSpeed}x${LinkWidth}                 \1/" >> /tmp/pcidevices2_link_sed.txt
		fi
	fi
	echo "/(# +$pcidevice \[....:....\] \[....)/s//\1${proginterface}/" >> /tmp/pcidevices1_progintf_sed.txt
done

perl -CSDA -Mutf8 -pe '
s/# (g.x..?) > \1 /# \1     /;     # if the links are the same then output it only once
s|(/s//#.{16}) *|\1|;              # make the link column 16 characters wide
' /tmp/pcidevices2_link_sed.txt > /tmp/pcidevices3_link_sed.txt

# If there's only one domain/segment then remove them from the output.
if [[ $(sed -E '/(....).*/s//\1/' /tmp/pcitree2.txt | sort -u | wc -l) -eq 1 ]]; then
	echo "/(#.{16}).{4}:/s//\\1/" >> /tmp/pcidevices3_link_sed.txt
fi

if [[ $fulladdress -eq 1 ]]; then
	echo "/..\..([^#]*#.{16})([0-9a-f.:]+) /s//\\2\\1/" >> /tmp/pcidevices3_link_sed.txt
fi


sed -E -f /tmp/pcidevices1_lspci_sed.txt -f /tmp/pcidevices1_progintf_sed.txt -f /tmp/pcidevices3_link_sed.txt /tmp/pcitree5.txt > /tmp/pcitree6.txt # apply the substitutions

echo '#========================================================================================='
cat /tmp/pcitree6.txt # final output
	#!/bin/bash
	# by joevt Jan 22, 2023

	#===================
	if [[ $EUID -ne 0 ]]; then
	echo "This script must be run as root like this:"
	echo "sudo $0"
	exit 1
	fi
	#===================
	style="wide"
	style="narrow"

	#encoding="ascii"
	encoding="utf8"

	#method="-H1" # -A intel-conf1
	method="" # default method which for macOS is -A darwin

	#===================
	if [ $encoding == "ascii" ]; then
	box0='\'
	box1='-'
	box2='\'
	box3='+'
	box4='\|'
	box5='.'
	else
	box0='┐'
	box1='─'
	box2='└'
	box3='├'
	box4='│'
	box5='┬'
	fi

	if [ $style == "wide" ]; then

	if [ $encoding == "ascii" ]; then
	# since we're using \, we want to indent one extra space
	rootindent=' '
	subindent=' '
	else
	rootindent=' '
	subindent=' '
	fi

	domainprefix="\\${box1}\\["
	domainsuffix="\\]\\${box1}\\${box0}"

	bridgeprefix="\\${box3}\\${box1}"
	bridgesuffixlong="\\${box1}\\${box1}\\${box1}\\${box1}\\${box0}"
	bridgesuffix="\\${box1}\\${box0}"

	deviceprefix="\\${box3}\\${box1}"
	devicesuffix=""
	else
	rootindent=''
	subindent=' '

	domainprefix="\\${box5}\\["
	domainsuffix="\\]"

	bridgeprefix="\\${box3}\\${box5}"
	bridgesuffixlong=""
	bridgesuffix=""

	deviceprefix="\\${box3}\\${box1}"
	devicesuffix=""

	fulladdress=1
	fi

	#===================

	# Get header type and bus information for all pci devices - reverse sort so last bus is first (required for device path substitution step below).
	# Format is "domain:bus%device.function domain:primary domain:secondary domain:subordinate".
	# Keep the domain number for device path substitution step below.
	# -0777 allows perl to process all lines at once - here we process two lines at a time.
	setpci $method -v -s '::*' HEADER_TYPE.b 0x18.l 2> /dev/null \| perl -0777 -pe '
	s/((.{4}):(..):(..\..)) \@0e = ([08][12])\n\1 \@18 = ..(.{2})(.{2})(.{2})\n/$2:$3%$4 $2:$8 $2:$7 $2:$6\n/g; # bridge(01) or card bus(02) header type, multifunction(80) or not (00).
	s/((.{4}):(..):(..\..)) \@0e = (..)\n\1 \@18 = ..(.{2})(.{2})(.{2})\n/$2:$3%$4\n/g # endpoint (or anything else) header type
	' \
	\| sort -r > /tmp/pcilist.txt

	# If primary = secondary = subordinate then that is a problem - change them to XX.
	sed '/^\(....:..%..\..\) \(\(.\{4\}:\)..\) \2 \2$/s//\1 \3XX \3XX \3XX/' /tmp/pcilist.txt > /tmp/pcilist2.txt

	echo '/ ....:/s// /g' > /tmp/pcipaths.txt # Add a command that will remove domain from primary, secondary, and subordinate bus numbers.

	# Create device path substitution commands that will add parent bus to each secondary bus.
	# When applied to the list of pci devices, the commands will create a full device path for
	# each pci device. Multiple commands will be applied to each device to build the device
	# path from right to left: end point (greatest bus number) to domain (lowest bus number).
	# The commands are sorted from highest bus number to lowest bus number to make that possible.
	sed -nE '/^(....:..%..\..) .{4}:.. (.{4}:..) .*/s//\/(\2)\/s\/\/\1%\\1\/g/p' /tmp/pcilist2.txt >> /tmp/pcipaths.txt
	sed -E -f /tmp/pcipaths.txt /tmp/pcilist2.txt > /tmp/pcitree1.txt # Apply the device path substitution commands.
	sed -E '/^(....:..).*/s//\1/g' /tmp/pcitree1.txt \| sort -u > /tmp/pcitree2.txt # Get all root complexes (usually just 0000:00).

	cat /tmp/pcitree1.txt /tmp/pcitree2.txt \| LC_ALL=C sort > /tmp/pcitree3.txt # Concatenate the devices and root complexes and sort.

	# Use perl to do some formatting.
	perl -CSDA -Mutf8 -pe '
	s/^(.(\w\w\w\w:\w\w%\w\w\.\w).)/\1 # \2/; # save domain/bus/device/function after a " # " - this will be used to append name and ids
	s/^\w\w\w\w:\w\w%/'"$rootindent"'/; # remove domain/bus number from left side and indent
	s/\w\w\.\w%\w\w\w\w:\w\w%/'"$subindent"'/g; # use the paths to do more indenting
	s/(\w\w\.\w) \w\w (\w\w) \2/'"${bridgeprefix}"'\1-[\2]'"${bridgesuffixlong}"'/; # format the secondary and subordinate bus numbers for single bus bridge
	s/(\w\w\.\w) \w\w (\w\w) (\w\w)/'"${bridgeprefix}"'\1-[\2-\3]'"${bridgesuffix}"'/; # format the secondary and subordinate bus numbers for multiple bus bridge
	s/%/:/; # replace % bus/device delimiter with standard ":"
	s/(\w\w\.\w )/'"${deviceprefix}"'\1'"${devicesuffix}"'/; # add the "+-" formating for devices
	s/^(\w\w\w\w:\w\w)$/'"${domainprefix}"'\1'"${domainsuffix}"'/ # add the -[ ]\" for domain/bus
	' \
	/tmp/pcitree3.txt > /tmp/pcitree4.txt

	# https://perldoc.perl.org/perlrun.html
	# The -C flag controls some of the Perl Unicode features.
	# S: STDIN, STDOUT and STDERR are assumed to be in UTF-8.
	# D: UTF-8 is the default PerlIO layer for both input and output streams.
	# A: @ARGV elements are expected to be strings encoded in UTF-8.
	# -0[octal/hexadecimal] specifies the input record separator ($/ ) as an octal or hexadecimal number.
	# Any value 0400 or above will cause Perl to slurp files whole, but by convention the value 0777 is the one normally used for this purpose.
	# -Mmodule executes use module ; before executing your program. This loads the module and calls its import method, causing the module to have its default effect, typically importing subroutines or giving effect to a pragma.
	# -p causes Perl to assume a loop around your program, which makes it iterate over filename arguments somewhat like sed.
	# -e commandline May be used to enter one line of program.
	perl -CSDA -Mutf8 -0777 -pe '

	# add \| between devices of the same bus
	do {
	$good=0;
	while (
	/ # begin regular expression
	( # begin g1
	\n # \n
	([\ '"\\${box4}"']) # [ \|] -> g2
	['"\\${box3}\\${box4}"'] # [+\|]
	[^\n]* # rest of the line up to \n, (change + to *+ to give nothing back)
	(?{$X=pos()}) # mark position
	\n # \n
	) # end g1
	(?= # begin lookahead
	( # begin group
	\2\ [^\n]* # following line(s) start with g2 also (change + to *+ to give nothing back)
	\n #
	)+ # end group, match one or more, (change + to ++ to give nothing back)
	\2 # the line after those start with g2
	'"\\${box3}"' # and "+" meaning that another device follows g1 on the same bus
	) # end lookahead
	(
	\2 #
	(?{$C=pos()}) # $C is the position where we want to insert the \|
	\ # the character at position $C to be modified is a space character
	) #
	/gx # end regular expression (g=global; x=ignore white space in RE)
	) {
	$good = 1;
	substr($_,$C,1) = "'"\\${box4}"'"; # replace the space character at position $C with a \|
	pos() = $X; # start next search at $X which is the end of the previous line
	}
	} until ($good == 0);

	# replace + with \ for last device of a bus
	s/(\n[\ '"\\${box4}"'])'"\\${box3}"'(?![^\n]\1['"\\${box3}\\${box4}"'])/\1'"\\${box2}"'/g;
	' /tmp/pcitree4.txt > /tmp/pcitree5.txt

	# Get the length of the longest line so we can align device information in a detail column.
	spaces=$(sed -E 's/./ /g' /tmp/pcitree5.txt \| sort -r \| head -n 1)


	lspci $method -nnv -D 2> /dev/null > /tmp/pcidevices0_lspci.txt
	perl -pe '
	s/^\n//g;
	s/^[ \t].*\n//g;
	s\|/\|\\/\|g;
	s\|^(....:..:..\..) ([^[]) (\[....\])(.) (\[....:....\])( \(rev ..\))?(?: \(prog-if ..( \[.*?\])\))?\|/# (\1)\$/s//'" "'# \\1 \5 \3:\6 : \2\7 \4/\|;
	s/\]:([^:]{9}.?) :([^:]{37}.?) :/]\1\2 :/ # rev column is 9 characters and type column is 36 characters
	' \
	/tmp/pcidevices0_lspci.txt > /tmp/pcidevices1_lspci_sed.txt
	echo '/^(.{'"$((${#spaces} - 14))"'}) (.)/s//\1\2/' >> /tmp/pcidevices1_lspci_sed.txt # create the substitutions for the detail column


	# Get link width and speed max/current.
	pat="(....:..:..\..) @09 = (..)"
	IFS=$'\n'
	echo "" > /tmp/pcidevices2_link_sed.txt
	echo "" > /tmp/pcidevices1_progintf_sed.txt
	for thedevice in $(sudo setpci $method -v -s '::*' CLASS_PROG.b 2> /dev/null); do
	pcidevice="${thedevice% @*}"
	proginterface="${thedevice#* = }"
	# We have to do this in a loop because setpci stops if express capability doesn't exist
	LinkRegisters=$(sudo setpci $method -s $pcidevice CAP_EXP+12.w CAP_EXP+c.l 2> /dev/null)
	if [[ -n $LinkRegisters ]]; then
	LinkStatusRegister=${LinkRegisters:0:4}
	LinkWidth=$((0x$LinkStatusRegister >> 4 & 31))
	if [[ $LinkWidth -ge 0 ]]; then
	LinkSpeed=$((0x$LinkStatusRegister & 15))
	LinkCapabilitiesRegister=${LinkRegisters:5:8}
	MaxLinkWidth=$((0x$LinkCapabilitiesRegister >> 4 & 31))
	MaxLinkSpeed=$((0x$LinkCapabilitiesRegister & 15))
	echo "/# ($pcidevice .*$)/s//# g${MaxLinkSpeed}x${MaxLinkWidth} > g${LinkSpeed}x${LinkWidth} \1/" >> /tmp/pcidevices2_link_sed.txt
	fi
	fi
	echo "/(# +$pcidevice \[....:....\] \[....)/s//\1${proginterface}/" >> /tmp/pcidevices1_progintf_sed.txt
	done

	perl -CSDA -Mutf8 -pe '
	s/# (g.x..?) > \1 /# \1 /; # if the links are the same then output it only once
	s\|(/s//#.{16}) *\|\1\|; # make the link column 16 characters wide
	' /tmp/pcidevices2_link_sed.txt > /tmp/pcidevices3_link_sed.txt

	# If there's only one domain/segment then remove them from the output.
	if [[ $(sed -E '/(....).*/s//\1/' /tmp/pcitree2.txt \| sort -u \| wc -l) -eq 1 ]]; then
	echo "/(#.{16}).{4}:/s//\\1/" >> /tmp/pcidevices3_link_sed.txt
	fi

	if [[ $fulladdress -eq 1 ]]; then
	echo "/..\..([^#]*#.{16})([0-9a-f.:]+) /s//\\2\\1/" >> /tmp/pcidevices3_link_sed.txt
	fi


	sed -E -f /tmp/pcidevices1_lspci_sed.txt -f /tmp/pcidevices1_progintf_sed.txt -f /tmp/pcidevices3_link_sed.txt /tmp/pcitree5.txt > /tmp/pcitree6.txt # apply the substitutions

	echo '#========================================================================================='
	cat /tmp/pcitree6.txt # final output