[Buildroot] [PATCH] support/scripts/cpedb.py: drop CPE XML database caching

[Buildroot] [PATCH] support/scripts/cpedb.py: drop CPE XML database caching

[Thomas Petazzoni]

Currently, the CPE XML database is parsed into a Python dict, which is
then pickled into a local file, to speed up the processing of further
invocations.

However, it turns out that since the initial implementation, we have
switched the XML parsing from the out of tree xmltodict module to the
standard ElementTree one, which has made the parsing much faster. The
pickle caching only saves 6 seconds, on something that takes more than
13 minutes total.

In addition, this pickle caching consumes a significant amount of RAM,
causing the Python process to be OOM-killed on a server with 4 GB of
RAM.

So let's just drop this caching entirely.

Signed-off-by: Thomas Petazzoni <thomas.petazzoni@bootlin.com>
---
This should be applied to master and next. Indeed, the pkg-stats
results used for autobuild.buildroot.org/stats/ are currently done on
next, but we also probably want people to have this change in master
for the 2021.02 release.
---
 support/scripts/cpedb.py | 40 ++++++----------------------------------
 1 file changed, 6 insertions(+), 34 deletions(-)

diff --git a/support/scripts/cpedb.py b/support/scripts/cpedb.py
index 825ed6cb1e..b1e7e7012c 100644
--- a/support/scripts/cpedb.py
+++ b/support/scripts/cpedb.py
@@ -94,22 +94,6 @@ class CPEDB:
         self.all_cpes_no_version = dict()
         self.nvd_path = nvd_path
 
-    def gen_cached_cpedb(self, cpedb, cache_all_cpes, cache_all_cpes_no_version):
-        print("CPE: Unzipping xml manifest...")
-        nist_cpe_file = gzip.GzipFile(fileobj=open(cpedb, 'rb'))
-        print("CPE: Converting xml manifest to dict...")
-        tree = ET.parse(nist_cpe_file)
-        all_cpedb = tree.getroot()
-        self.parse_dict(all_cpedb)
-
-        print("CPE: Caching dictionary")
-        cpes_file = open(cache_all_cpes, 'wb')
-        pickle.dump(self.all_cpes, cpes_file)
-        cpes_file.close()
-        cpes_file = open(cache_all_cpes_no_version, 'wb')
-        pickle.dump(self.all_cpes_no_version, cpes_file)
-        cpes_file.close()
-
     def get_xml_dict(self):
         print("CPE: Setting up NIST dictionary")
         if not os.path.exists(os.path.join(self.nvd_path, "cpe")):
@@ -121,24 +105,12 @@ class CPEDB:
             cpe_dict = requests.get(CPEDB_URL)
             open(cpe_dict_local, "wb").write(cpe_dict.content)
 
-        cache_all_cpes = os.path.join(self.nvd_path, "cpe", "all_cpes.pkl")
-        cache_all_cpes_no_version = os.path.join(self.nvd_path, "cpe", "all_cpes_no_version.pkl")
-
-        if not os.path.exists(cache_all_cpes) or \
-           not os.path.exists(cache_all_cpes_no_version) or \
-           os.stat(cache_all_cpes).st_mtime < os.stat(cpe_dict_local).st_mtime or \
-           os.stat(cache_all_cpes_no_version).st_mtime < os.stat(cpe_dict_local).st_mtime:
-            self.gen_cached_cpedb(cpe_dict_local,
-                                  cache_all_cpes,
-                                  cache_all_cpes_no_version)
-
-        print("CPE: Loading CACHED dictionary")
-        cpe_file = open(cache_all_cpes, 'rb')
-        self.all_cpes = pickle.load(cpe_file)
-        cpe_file.close()
-        cpe_file = open(cache_all_cpes_no_version, 'rb')
-        self.all_cpes_no_version = pickle.load(cpe_file)
-        cpe_file.close()
+        print("CPE: Unzipping xml manifest...")
+        nist_cpe_file = gzip.GzipFile(fileobj=open(cpe_dict_local, 'rb'))
+        print("CPE: Converting xml manifest to dict...")
+        tree = ET.parse(nist_cpe_file)
+        all_cpedb = tree.getroot()
+        self.parse_dict(all_cpedb)
 
     def parse_dict(self, all_cpedb):
         # Cycle through the dict and build two dict to be used for custom
-- 
2.29.2

[Buildroot] [PATCH] support/scripts/cpedb.py: drop CPE XML database caching

[Yann E. MORIN]

Thomas, All,

On 2021-02-13 23:19 +0100, Thomas Petazzoni spake thusly:
> Currently, the CPE XML database is parsed into a Python dict, which is
> then pickled into a local file, to speed up the processing of further
> invocations.
> 
> However, it turns out that since the initial implementation, we have
> switched the XML parsing from the out of tree xmltodict module to the
> standard ElementTree one, which has made the parsing much faster. The
> pickle caching only saves 6 seconds, on something that takes more than
> 13 minutes total.
> 
> In addition, this pickle caching consumes a significant amount of RAM,
> causing the Python process to be OOM-killed on a server with 4 GB of
> RAM.
> 
> So let's just drop this caching entirely.
> 
> Signed-off-by: Thomas Petazzoni <thomas.petazzoni@bootlin.com>
> ---
> This should be applied to master and next. Indeed, the pkg-stats
> results used for autobuild.buildroot.org/stats/ are currently done on
> next, but we also probably want people to have this change in master
> for the 2021.02 release.

Applied to master and next, thanks.

Note a comment below...

> ---
>  support/scripts/cpedb.py | 40 ++++++----------------------------------
>  1 file changed, 6 insertions(+), 34 deletions(-)
> 
> diff --git a/support/scripts/cpedb.py b/support/scripts/cpedb.py
> index 825ed6cb1e..b1e7e7012c 100644
> --- a/support/scripts/cpedb.py
> +++ b/support/scripts/cpedb.py
[--SNIP--]
> @@ -121,24 +105,12 @@ class CPEDB:
>              cpe_dict = requests.get(CPEDB_URL)
>              open(cpe_dict_local, "wb").write(cpe_dict.content)
>  
> -        cache_all_cpes = os.path.join(self.nvd_path, "cpe", "all_cpes.pkl")
> -        cache_all_cpes_no_version = os.path.join(self.nvd_path, "cpe", "all_cpes_no_version.pkl")
> -
> -        if not os.path.exists(cache_all_cpes) or \
> -           not os.path.exists(cache_all_cpes_no_version) or \
> -           os.stat(cache_all_cpes).st_mtime < os.stat(cpe_dict_local).st_mtime or \
> -           os.stat(cache_all_cpes_no_version).st_mtime < os.stat(cpe_dict_local).st_mtime:
> -            self.gen_cached_cpedb(cpe_dict_local,
> -                                  cache_all_cpes,
> -                                  cache_all_cpes_no_version)
> -
> -        print("CPE: Loading CACHED dictionary")
> -        cpe_file = open(cache_all_cpes, 'rb')
> -        self.all_cpes = pickle.load(cpe_file)
> -        cpe_file.close()
> -        cpe_file = open(cache_all_cpes_no_version, 'rb')
> -        self.all_cpes_no_version = pickle.load(cpe_file)
> -        cpe_file.close()
> +        print("CPE: Unzipping xml manifest...")
> +        nist_cpe_file = gzip.GzipFile(fileobj=open(cpe_dict_local, 'rb'))
> +        print("CPE: Converting xml manifest to dict...")
> +        tree = ET.parse(nist_cpe_file)

Once your nist_cpe_file has been parsed, you could delete it to reclaim
some memory:

    del nist_cpe_file

And maybe do so for a few other intemediate blobs that are really big...

Regards,
Yann E. MORIN.

> +        all_cpedb = tree.getroot()
> +        self.parse_dict(all_cpedb)
>  
>      def parse_dict(self, all_cpedb):
>          # Cycle through the dict and build two dict to be used for custom
> -- 
> 2.29.2
> 

-- 
.-----------------.--------------------.------------------.--------------------.
|  Yann E. MORIN  | Real-Time Embedded | /"\ ASCII RIBBON | Erics' conspiracy: |
| +33 662 376 056 | Software  Designer | \ / CAMPAIGN     |  ___               |
| +33 561 099 427 `------------.-------:  X  AGAINST      |  \e/  There is no  |
| http://ymorin.is-a-geek.org/ | _/*\_ | / \ HTML MAIL    |   v   conspiracy.  |
'------------------------------^-------^------------------^--------------------'