Patchwork [RFC,v2,06/11] pstore: Add decompression support to pstore

login
register
mail settings
Submitter Aruna Balakrishnaiah
Date Aug. 16, 2013, 1:18 p.m.
Message ID <20130816131821.3338.27923.stgit@aruna-ThinkPad-T420>
Download mbox | patch
Permalink /patch/267645/
State Superseded
Headers show

Comments

Aruna Balakrishnaiah - Aug. 16, 2013, 1:18 p.m.
Based on the flag 'compressed' set or not, pstore will decompress the
data returning a plain text file. If decompression fails for a particular
record it will have the compressed data in the file which can be
decompressed with 'openssl' command line tool.

Signed-off-by: Aruna Balakrishnaiah <aruna@linux.vnet.ibm.com>
---
 fs/pstore/platform.c |   53 ++++++++++++++++++++++++++++++++++++++++++++++++--
 1 file changed, 51 insertions(+), 2 deletions(-)
Seiji Aguchi - Aug. 22, 2013, 11:04 p.m.
> -----Original Message-----
> From: linux-kernel-owner@vger.kernel.org [mailto:linux-kernel-owner@vger.kernel.org] On Behalf Of Aruna Balakrishnaiah
> Sent: Friday, August 16, 2013 9:18 AM
> To: linuxppc-dev@ozlabs.org; tony.luck@intel.com; linux-kernel@vger.kernel.org; keescook@chromium.org
> Cc: jkenisto@linux.vnet.ibm.com; ananth@in.ibm.com; benh@kernel.crashing.org; cbouatmailru@gmail.com;
> mahesh@linux.vnet.ibm.com; ccross@android.com
> Subject: [RFC PATCH v2 06/11] pstore: Add decompression support to pstore
> 
> Based on the flag 'compressed' set or not, pstore will decompress the
> data returning a plain text file. If decompression fails for a particular
> record it will have the compressed data in the file which can be
> decompressed with 'openssl' command line tool.

If the decompression fails and openssl doesn't work, the worst case is that users can't read the entry.
In that case, pstore is meaningless at all.

Also, for users who want to get a single panic message, a compression is not needed.

So, I think we still have to support non-compression mode.
(IMO, pstore can take kdump as a model. Kdump supports both compression and non-compression mode.)

But, if you think my comment is outside this patchset, it's OK.
We can make it with a separate patch.

Seiji
Aruna Balakrishnaiah - Aug. 27, 2013, 9:39 a.m.
On Friday 23 August 2013 04:34 AM, Seiji Aguchi wrote:
>
>> -----Original Message-----
>> From: linux-kernel-owner@vger.kernel.org [mailto:linux-kernel-owner@vger.kernel.org] On Behalf Of Aruna Balakrishnaiah
>> Sent: Friday, August 16, 2013 9:18 AM
>> To: linuxppc-dev@ozlabs.org; tony.luck@intel.com; linux-kernel@vger.kernel.org; keescook@chromium.org
>> Cc: jkenisto@linux.vnet.ibm.com; ananth@in.ibm.com; benh@kernel.crashing.org; cbouatmailru@gmail.com;
>> mahesh@linux.vnet.ibm.com; ccross@android.com
>> Subject: [RFC PATCH v2 06/11] pstore: Add decompression support to pstore
>>
>> Based on the flag 'compressed' set or not, pstore will decompress the
>> data returning a plain text file. If decompression fails for a particular
>> record it will have the compressed data in the file which can be
>> decompressed with 'openssl' command line tool.
> If the decompression fails and openssl doesn't work, the worst case is that users can't read the entry.
> In that case, pstore is meaningless at all.

If decompression fails and openssl doesn't work. We have python module zlib to 
decompress
the zlib data. zlib.decompress() should do the trick.

> Also, for users who want to get a single panic message, a compression is not needed.
>
> So, I think we still have to support non-compression mode.
> (IMO, pstore can take kdump as a model. Kdump supports both compression and non-compression mode.)
>
> But, if you think my comment is outside this patchset, it's OK.
> We can make it with a separate patch.
>
> Seiji
> _______________________________________________
> Linuxppc-dev mailing list
> Linuxppc-dev@lists.ozlabs.org
> https://lists.ozlabs.org/listinfo/linuxppc-dev
>

Patch

diff --git a/fs/pstore/platform.c b/fs/pstore/platform.c
index 6418eb7..0195cca0 100644
--- a/fs/pstore/platform.c
+++ b/fs/pstore/platform.c
@@ -162,6 +162,36 @@  error:
 	return ret;
 }
 
+/* Derived from logfs_uncompress */
+int pstore_decompress(void *in, void *out, size_t inlen, size_t outlen)
+{
+	int err, ret;
+
+	ret = -EIO;
+	err = zlib_inflateInit(&stream);
+	if (err != Z_OK)
+		goto error;
+
+	stream.next_in = in;
+	stream.avail_in = inlen;
+	stream.total_in = 0;
+	stream.next_out = out;
+	stream.avail_out = outlen;
+	stream.total_out = 0;
+
+	err = zlib_inflate(&stream, Z_FINISH);
+	if (err != Z_STREAM_END)
+		goto error;
+
+	err = zlib_inflateEnd(&stream);
+	if (err != Z_OK)
+		goto error;
+
+	ret = stream.total_out;
+error:
+	return ret;
+}
+
 static void allocate_buf_for_compression(void)
 {
 	size_t size;
@@ -429,6 +459,7 @@  void pstore_get_records(int quiet)
 	struct timespec		time;
 	int			failed = 0, rc;
 	bool			compressed;
+	int			unzipped_len = -1;
 
 	if (!psi)
 		return;
@@ -439,10 +470,28 @@  void pstore_get_records(int quiet)
 
 	while ((size = psi->read(&id, &type, &count, &time, &buf, &compressed,
 				psi)) > 0) {
+		if (compressed && (type == PSTORE_TYPE_DMESG)) {
+			if (big_oops_buf)
+				unzipped_len = pstore_decompress(buf,
+							big_oops_buf, size,
+							big_oops_buf_sz);
+
+			if (unzipped_len > 0) {
+				buf = big_oops_buf;
+				size = unzipped_len;
+			} else {
+				pr_err("pstore: decompression failed;"
+					"returned %d\n", unzipped_len);
+			}
+		}
 		rc = pstore_mkfile(type, psi->name, id, count, buf,
 				  (size_t)size, time, psi);
-		kfree(buf);
-		buf = NULL;
+		if (unzipped_len < 0) {
+			/* Free buffer other than big oops */
+			kfree(buf);
+			buf = NULL;
+		} else
+			unzipped_len = -1;
 		if (rc && (rc != -EEXIST || !quiet))
 			failed++;
 	}