Hi Greg,
On Oct 15 '15 10:37, Valentin Rothberg wrote:
Can you take the patch below? I use the patched version of the script
since sending it and it has proved to be much faster and hence more
comfortable to use since then.
Kind regards,
Valentin
> Distribute the parsing of source and Kconfig files on all available
> cores to speed up processing.
>
> Signed-off-by: Valentin Rothberg <[email protected]>
> ---
> scripts/checkkconfigsymbols.py | 128 ++++++++++++++++++++++++++++++++---------
> 1 file changed, 102 insertions(+), 26 deletions(-)
>
> diff --git a/scripts/checkkconfigsymbols.py b/scripts/checkkconfigsymbols.py
> index 2f4b7ffd5570..cfe397b61c48 100755
> --- a/scripts/checkkconfigsymbols.py
> +++ b/scripts/checkkconfigsymbols.py
> @@ -10,9 +10,11 @@
>
> import os
> import re
> +import signal
> import sys
> -from subprocess import Popen, PIPE, STDOUT
> +from multiprocessing import Pool, cpu_count
> from optparse import OptionParser
> +from subprocess import Popen, PIPE, STDOUT
>
>
> # regex expressions
> @@ -26,7 +28,7 @@ SOURCE_FEATURE = r"(?:\W|\b)+[D]{,1}CONFIG_(" + FEATURE + r")"
>
> # regex objects
> REGEX_FILE_KCONFIG = re.compile(r".*Kconfig[\.\w+\-]*$")
> -REGEX_FEATURE = re.compile(r'(?!\B"[^"]*)' + FEATURE + r'(?![^"]*"\B)')
> +REGEX_FEATURE = re.compile(r'(?!\B)' + FEATURE + r'(?!\B)')
> REGEX_SOURCE_FEATURE = re.compile(SOURCE_FEATURE)
> REGEX_KCONFIG_DEF = re.compile(DEF)
> REGEX_KCONFIG_EXPR = re.compile(EXPR)
> @@ -34,6 +36,7 @@ REGEX_KCONFIG_STMT = re.compile(STMT)
> REGEX_KCONFIG_HELP = re.compile(r"^\s+(help|---help---)\s*$")
> REGEX_FILTER_FEATURES = re.compile(r"[A-Za-z0-9]$")
> REGEX_NUMERIC = re.compile(r"0[xX][0-9a-fA-F]+|[0-9]+")
> +REGEX_QUOTES = re.compile("(\"(.*?)\")")
>
>
> def parse_options():
> @@ -209,14 +212,36 @@ def get_head():
> return stdout.strip('\n')
>
>
> +def partition(lst, size):
> + """Partition list @lst into eveni-sized lists of size @size."""
> + return [lst[i::size] for i in xrange(size)]
> +
> +
> +def init_worker():
> + """Set signal handler to ignore SIGINT."""
> + signal.signal(signal.SIGINT, signal.SIG_IGN)
> +
> +
> def check_symbols(ignore):
> """Find undefined Kconfig symbols and return a dict with the symbol as key
> and a list of referencing files as value. Files matching %ignore are not
> checked for undefined symbols."""
> + pool = Pool(cpu_count(), init_worker)
> + try:
> + return check_symbols_helper(pool, ignore)
> + except KeyboardInterrupt:
> + pool.terminate()
> + pool.join()
> + sys.exit(1)
> +
> +
> +def check_symbols_helper(pool, ignore):
> + """Helper method for check_symbols(). Used to catch keyboard interrupts in
> + check_symbols() in order to properly terminate running worker processes."""
> source_files = []
> kconfig_files = []
> - defined_features = set()
> - referenced_features = dict() # {feature: [files]}
> + defined_features = []
> + referenced_features = dict() # {file: [features]}
>
> # use 'git ls-files' to get the worklist
> stdout = execute("git ls-files")
> @@ -231,21 +256,33 @@ def check_symbols(ignore):
> if REGEX_FILE_KCONFIG.match(gitfile):
> kconfig_files.append(gitfile)
> else:
> - # all non-Kconfig files are checked for consistency
> + if ignore and not re.match(ignore, gitfile):
> + continue
> + # add source files that do not match the ignore pattern
> source_files.append(gitfile)
>
> - for sfile in source_files:
> - if ignore and re.match(ignore, sfile):
> - # do not check files matching %ignore
> - continue
> - parse_source_file(sfile, referenced_features)
> + # parse source files
> + arglist = partition(source_files, cpu_count())
> + for res in pool.map(parse_source_files, arglist):
> + referenced_features.update(res)
>
> - for kfile in kconfig_files:
> - if ignore and re.match(ignore, kfile):
> - # do not collect references for files matching %ignore
> - parse_kconfig_file(kfile, defined_features, dict())
> - else:
> - parse_kconfig_file(kfile, defined_features, referenced_features)
> +
> + # parse kconfig files
> + arglist = []
> + for part in partition(kconfig_files, cpu_count()):
> + arglist.append((part, ignore))
> + for res in pool.map(parse_kconfig_files, arglist):
> + defined_features.extend(res[0])
> + referenced_features.update(res[1])
> + defined_features = set(defined_features)
> +
> + # inverse mapping of referenced_features to dict(feature: [files])
> + inv_map = dict()
> + for _file, features in referenced_features.iteritems():
> + for feature in features:
> + inv_map[feature] = inv_map.get(feature, set())
> + inv_map[feature].add(_file)
> + referenced_features = inv_map
>
> undefined = {} # {feature: [files]}
> for feature in sorted(referenced_features):
> @@ -262,9 +299,23 @@ def check_symbols(ignore):
> return undefined
>
>
> -def parse_source_file(sfile, referenced_features):
> - """Parse @sfile for referenced Kconfig features."""
> +def parse_source_files(source_files):
> + """Parse each source file in @source_files and return dictionary with source
> + files as keys and lists of references Kconfig symbols as values."""
> + referenced_features = dict()
> + for sfile in source_files:
> + referenced_features[sfile] = parse_source_file(sfile)
> + return referenced_features
> +
> +
> +def parse_source_file(sfile):
> + """Parse @sfile and return a list of referenced Kconfig features."""
> lines = []
> + references = []
> +
> + if not os.path.exists(sfile):
> + return references
> +
> with open(sfile, "r") as stream:
> lines = stream.readlines()
>
> @@ -275,9 +326,9 @@ def parse_source_file(sfile, referenced_features):
> for feature in features:
> if not REGEX_FILTER_FEATURES.search(feature):
> continue
> - sfiles = referenced_features.get(feature, set())
> - sfiles.add(sfile)
> - referenced_features[feature] = sfiles
> + references.append(feature)
> +
> + return references
>
>
> def get_features_in_line(line):
> @@ -285,11 +336,35 @@ def get_features_in_line(line):
> return REGEX_FEATURE.findall(line)
>
>
> -def parse_kconfig_file(kfile, defined_features, referenced_features):
> +def parse_kconfig_files(args):
> + """Parse kconfig files and return tuple of defined and references Kconfig
> + symbols. Note, @args is a tuple of a list of files and the @ignore
> + pattern."""
> + kconfig_files = args[0]
> + ignore = args[1]
> + defined_features = []
> + referenced_features = dict()
> +
> + for kfile in kconfig_files:
> + defined, references = parse_kconfig_file(kfile)
> + defined_features.extend(defined)
> + if ignore and re.match(ignore, kfile):
> + # do not collect references for files that match the ignore pattern
> + continue
> + referenced_features[kfile] = references
> + return (defined_features, referenced_features)
> +
> +
> +def parse_kconfig_file(kfile):
> """Parse @kfile and update feature definitions and references."""
> lines = []
> + defined = []
> + references = []
> skip = False
>
> + if not os.path.exists(kfile):
> + return defined, references
> +
> with open(kfile, "r") as stream:
> lines = stream.readlines()
>
> @@ -300,7 +375,7 @@ def parse_kconfig_file(kfile, defined_features, referenced_features):
>
> if REGEX_KCONFIG_DEF.match(line):
> feature_def = REGEX_KCONFIG_DEF.findall(line)
> - defined_features.add(feature_def[0])
> + defined.append(feature_def[0])
> skip = False
> elif REGEX_KCONFIG_HELP.match(line):
> skip = True
> @@ -308,6 +383,7 @@ def parse_kconfig_file(kfile, defined_features, referenced_features):
> # ignore content of help messages
> pass
> elif REGEX_KCONFIG_STMT.match(line):
> + line = REGEX_QUOTES.sub("", line)
> features = get_features_in_line(line)
> # multi-line statements
> while line.endswith("\\"):
> @@ -319,9 +395,9 @@ def parse_kconfig_file(kfile, defined_features, referenced_features):
> if REGEX_NUMERIC.match(feature):
> # ignore numeric values
> continue
> - paths = referenced_features.get(feature, set())
> - paths.add(kfile)
> - referenced_features[feature] = paths
> + references.append(feature)
> +
> + return defined, references
>
>
> if __name__ == "__main__":
> --
> 1.9.1
>
On Sun, Nov 01, 2015 at 01:53:40PM +0100, Valentin Rothberg wrote:
> Hi Greg,
>
> On Oct 15 '15 10:37, Valentin Rothberg wrote:
>
> Can you take the patch below? I use the patched version of the script
> since sending it and it has proved to be much faster and hence more
> comfortable to use since then.
It's a bit late in the merge window cycle, sorry, can you resend this
after 4.4-rc1 is out and I will be glad to take it.
I didn't see this in my patch queue, did you not send it to me
previously?
thanks,
greg k-h
On Nov 01 '15 10:13, Greg KH wrote:
> On Sun, Nov 01, 2015 at 01:53:40PM +0100, Valentin Rothberg wrote:
> > Hi Greg,
> >
> > On Oct 15 '15 10:37, Valentin Rothberg wrote:
> >
> > Can you take the patch below? I use the patched version of the script
> > since sending it and it has proved to be much faster and hence more
> > comfortable to use since then.
>
> It's a bit late in the merge window cycle, sorry, can you resend this
> after 4.4-rc1 is out and I will be glad to take it.
Sure, with pleasure :)
> I didn't see this in my patch queue, did you not send it to me
> previously?
I sent it directly to you. Did you mark me as spam? Just kidding.
Maybe the mail just got lost in your mailbox since the script can't be
assigned to a particular subsystem.
Best regards,
Valentin
On Sun, Nov 01, 2015 at 07:30:14PM +0100, Valentin Rothberg wrote:
> On Nov 01 '15 10:13, Greg KH wrote:
> > On Sun, Nov 01, 2015 at 01:53:40PM +0100, Valentin Rothberg wrote:
> > > Hi Greg,
> > >
> > > On Oct 15 '15 10:37, Valentin Rothberg wrote:
> > >
> > > Can you take the patch below? I use the patched version of the script
> > > since sending it and it has proved to be much faster and hence more
> > > comfortable to use since then.
> >
> > It's a bit late in the merge window cycle, sorry, can you resend this
> > after 4.4-rc1 is out and I will be glad to take it.
>
> Sure, with pleasure :)
>
> > I didn't see this in my patch queue, did you not send it to me
> > previously?
>
> I sent it directly to you. Did you mark me as spam? Just kidding.
> Maybe the mail just got lost in your mailbox since the script can't be
> assigned to a particular subsystem.
When did you send it? Any message id I can search for?
thanks,
greg k-h
On Nov 01 '15 11:45, Greg KH wrote:
> On Sun, Nov 01, 2015 at 07:30:14PM +0100, Valentin Rothberg wrote:
> > On Nov 01 '15 10:13, Greg KH wrote:
> > > On Sun, Nov 01, 2015 at 01:53:40PM +0100, Valentin Rothberg wrote:
> > > > Hi Greg,
> > > >
> > > > On Oct 15 '15 10:37, Valentin Rothberg wrote:
> > > >
> > > > Can you take the patch below? I use the patched version of the script
> > > > since sending it and it has proved to be much faster and hence more
> > > > comfortable to use since then.
> > >
> > > It's a bit late in the merge window cycle, sorry, can you resend this
> > > after 4.4-rc1 is out and I will be glad to take it.
> >
> > Sure, with pleasure :)
> >
> > > I didn't see this in my patch queue, did you not send it to me
> > > previously?
> >
> > I sent it directly to you. Did you mark me as spam? Just kidding.
> > Maybe the mail just got lost in your mailbox since the script can't be
> > assigned to a particular subsystem.
>
> When did you send it? Any message id I can search for?
Thanks for checking.
Subject: [PATCH] checkkconfigsymbols.py: multiprocessing of files
Date: Thu, 15 Oct 2015 10:37:47 +0200
Message-Id: <[email protected]>
Kind regards,
Valentin
On Sun, Nov 01, 2015 at 09:56:18PM +0100, Valentin Rothberg wrote:
> On Nov 01 '15 11:45, Greg KH wrote:
> > On Sun, Nov 01, 2015 at 07:30:14PM +0100, Valentin Rothberg wrote:
> > > On Nov 01 '15 10:13, Greg KH wrote:
> > > > On Sun, Nov 01, 2015 at 01:53:40PM +0100, Valentin Rothberg wrote:
> > > > > Hi Greg,
> > > > >
> > > > > On Oct 15 '15 10:37, Valentin Rothberg wrote:
> > > > >
> > > > > Can you take the patch below? I use the patched version of the script
> > > > > since sending it and it has proved to be much faster and hence more
> > > > > comfortable to use since then.
> > > >
> > > > It's a bit late in the merge window cycle, sorry, can you resend this
> > > > after 4.4-rc1 is out and I will be glad to take it.
> > >
> > > Sure, with pleasure :)
> > >
> > > > I didn't see this in my patch queue, did you not send it to me
> > > > previously?
> > >
> > > I sent it directly to you. Did you mark me as spam? Just kidding.
> > > Maybe the mail just got lost in your mailbox since the script can't be
> > > assigned to a particular subsystem.
> >
> > When did you send it? Any message id I can search for?
>
> Thanks for checking.
>
> Subject: [PATCH] checkkconfigsymbols.py: multiprocessing of files
> Date: Thu, 15 Oct 2015 10:37:47 +0200
> Message-Id: <[email protected]>
Turns out that Google had marked this as spam, so I didn't receive it on
my end, sorry about that. For 2 days it looks like Google got a bit
"spam happy" and started marking lots of things it shouldn't have as
spam, hopefully that should be fixed now.
I'll queue this up after 4.4-rc1 is out, sorry for the delay, not your
fault.
greg k-h
On Nov 01 '15 19:13, Greg KH wrote:
> On Sun, Nov 01, 2015 at 09:56:18PM +0100, Valentin Rothberg wrote:
> > On Nov 01 '15 11:45, Greg KH wrote:
> > > On Sun, Nov 01, 2015 at 07:30:14PM +0100, Valentin Rothberg wrote:
> > > > On Nov 01 '15 10:13, Greg KH wrote:
> > > > > On Sun, Nov 01, 2015 at 01:53:40PM +0100, Valentin Rothberg wrote:
> > > > > > Hi Greg,
> > > > > >
> > > > > > On Oct 15 '15 10:37, Valentin Rothberg wrote:
> > > > > >
> > > > > > Can you take the patch below? I use the patched version of the script
> > > > > > since sending it and it has proved to be much faster and hence more
> > > > > > comfortable to use since then.
> > > > >
> > > > > It's a bit late in the merge window cycle, sorry, can you resend this
> > > > > after 4.4-rc1 is out and I will be glad to take it.
> > > >
> > > > Sure, with pleasure :)
> > > >
> > > > > I didn't see this in my patch queue, did you not send it to me
> > > > > previously?
> > > >
> > > > I sent it directly to you. Did you mark me as spam? Just kidding.
> > > > Maybe the mail just got lost in your mailbox since the script can't be
> > > > assigned to a particular subsystem.
> > >
> > > When did you send it? Any message id I can search for?
> >
> > Thanks for checking.
> >
> > Subject: [PATCH] checkkconfigsymbols.py: multiprocessing of files
> > Date: Thu, 15 Oct 2015 10:37:47 +0200
> > Message-Id: <[email protected]>
>
> Turns out that Google had marked this as spam, so I didn't receive it on
> my end, sorry about that. For 2 days it looks like Google got a bit
> "spam happy" and started marking lots of things it shouldn't have as
> spam, hopefully that should be fixed now.
Oh, I remember having the same issue at that time. I 'solved' the issue
with Gmail filters avoiding to mark anything on lkml as spam, which in
turn means having more spam...
> I'll queue this up after 4.4-rc1 is out, sorry for the delay, not your
> fault.
Thank you.
Kind regards,
Valentin
Hi Greg,
On Nov 01 '15 19:13, Greg KH wrote:
[...]
> I'll queue this up after 4.4-rc1 is out, sorry for the delay, not your
> fault.
Can you queue the patch for checkkconfigsymbols.py? I would like to
send another patch adding support to find similar symbols, and apply it
ontop of the multi-processing patch.
Kind regards,
Valentin
On Fri, Nov 20, 2015 at 10:52:32AM +0100, Valentin Rothberg wrote:
> Hi Greg,
>
> On Nov 01 '15 19:13, Greg KH wrote:
> [...]
> > I'll queue this up after 4.4-rc1 is out, sorry for the delay, not your
> > fault.
>
> Can you queue the patch for checkkconfigsymbols.py? I would like to
> send another patch adding support to find similar symbols, and apply it
> ontop of the multi-processing patch.
It's in my very-full-queue :(
Feel free to send a patch based on this one, I will apply them in the
correct order.
thanks,
greg k-h