Skip to contentSkip to conversation

xen-unstable build failure in tools/blktap/drivers for block-qcow2.c

PATCH New Srinivas Maturi 3 participants 8 messages Feb 27, 2008 – Feb 29, 2008

lore
Share & tools
lore
#0 lore raw

We are seeing build failure of xen-unstable for the last couple of changesets, in tools/blktap/drivers for block-qcow2.c

This seems to be because of the fact that the compile line uses these flags, -Werror -Wno-unused, treating warnings as errors and aborting the compilation.

The Makefile in tools/blktap/drivers has these CFLAGS set:
CFLAGS += -Werror
CFLAGS += -Wno-unused

gcc -O2 -g -pipe -Wall -Wp,-D_FORTIFY_SOURCE=2 -fexceptions -fstack-protector --param=ssp-buffer-size=4 -m64 -mtune=generic -m64 -fno-strict-aliasing -std=gnu99 -Wall -Wstrict-prototypes -Wno-unused-value -Wdeclaration-after-statement -m64 -fno-strict-aliasing -std=gnu99 -Wall -Wstrict-prototypes -Wno-unused-value -Wdeclaration-after-statement -m64 -fno-strict-aliasing -std=gnu99 -Wall -Wstrict-prototypes -Wno-unused-value -Wdeclaration-after-statement -D__XEN_TOOLS__ -D_LARGEFILE_SOURCE -D_LARGEFILE64_SOURCE -m64 -fno-strict-aliasing -std=gnu99 -Wall -Wstrict-prototypes -Wno-unused-value -Wdeclaration-after-statement -D__XEN_TOOLS__ -D_LARGEFILE_SOURCE -D_LARGEFILE64_SOURCE -m64 -fno-strict-aliasing -std=gnu99 -Wall -Wstrict-prototypes -Wno-unused-value -Wdeclaration-after-statement -D__XEN_TOOLS__ -D_LARGEFILE_SOURCE -D_LARGEFILE64_SOURCE -m64 -fno-strict-aliasing -std=gnu99 -Wall -Wstrict-prototypes -Wno-unused-value -Wdeclaration-after-statement -D__XEN_TOOLS__ -D_LARGEFILE_SOURCE -D_LARGEFILE64_SOURCE -Werror -Wno-unused -I../../../tools/libxc -I../../../tools/include -I ../../libaio/src -I.. -I../lib -I../../../tools/include -I. -I../../xenstore -D_GNU_SOURCE -Wp,-MD,.block-qcow2.o.d -c -o block-qcow2.o block-qcow2.c
cc1: warnings being treated as errors
block-qcow2.c: In function ‘qcow2_create’:
block-qcow2.c:1312: warning: ignoring return value of ‘write’, declared with attribute warn_unused_result
block-qcow2.c:1314: warning: ignoring return value of ‘write’, declared with attribute warn_unused_result
block-qcow2.c:1319: warning: ignoring return value of ‘write’, declared with attribute warn_unused_result
block-qcow2.c:1322: warning: ignoring return value of ‘write’, declared with attribute warn_unused_result
block-qcow2.c:1325: warning: ignoring return value of ‘write’, declared with attribute warn_unused_result
block-qcow2.c: In function ‘qcow_write_compressed’:
block-qcow2.c:1354: warning: ignoring return value of ‘ftruncate’, declared with attribute warn_unused_result
.....
.....

Thanks,
Maturi

#1 lore raw

Which version of gcc are you using? It works for me with 3.4.4 and 4.1.1.

Properly speaking we should check those write() return values anyway. If we don't want to deviate from the qemu original sources that much then we could special-case CFLAGS to remove -Wno-unused for just that file.

-- Keir

Quoted 59 lines · Srinivas Maturi

We are seeing build failure of xen-unstable for the last couple of changesets,
in tools/blktap/drivers for block-qcow2.c

This seems to be because of the fact that the compile line uses these flags, -Werror -Wno-unused, treating warnings as errors and aborting the compilation.

The Makefile in tools/blktap/drivers has these CFLAGS set:
CFLAGS += -Werror
CFLAGS += -Wno-unused

gcc -O2 -g -pipe -Wall -Wp,-D_FORTIFY_SOURCE=2 -fexceptions

(quoted diff) +0 -19
-fstack-protector --param=ssp-buffer-size=4 -m64 -mtune=generic -m64
-fno-strict-aliasing -std=gnu99 -Wall -Wstrict-prototypes
-Wno-unused-value -Wdeclaration-after-statement -m64
-fno-strict-aliasing -std=gnu99 -Wall -Wstrict-prototypes
-Wno-unused-value -Wdeclaration-after-statement -m64
-fno-strict-aliasing -std=gnu99 -Wall -Wstrict-prototypes
-Wno-unused-value -Wdeclaration-after-statement -D__XEN_TOOLS__
-D_LARGEFILE_SOURCE -D_LARGEFILE64_SOURCE -m64 -fno-strict-aliasing
-std=gnu99 -Wall -Wstrict-prototypes -Wno-unused-value
-Wdeclaration-after-statement -D__XEN_TOOLS__ -D_LARGEFILE_SOURCE
-D_LARGEFILE64_SOURCE -m64 -fno-strict-aliasing -std=gnu99 -Wall
-Wstrict-prototypes -Wno-unused-value -Wdeclaration-after-statement
-D__XEN_TOOLS__ -D_LARGEFILE_SOURCE -D_LARGEFILE64_SOURCE -m64
-fno-strict-aliasing -std=gnu99 -Wall -Wstrict-prototypes
-Wno-unused-value -Wdeclaration-after-statement -D__XEN_TOOLS__
-D_LARGEFILE_SOURCE -D_LARGEFILE64_SOURCE -Werror -Wno-unused
-I../../../tools/libxc -I../../../tools/include -I ../../libaio/src -I..
-I../lib -I../../../tools/include -I. -I../../xenstore -D_GNU_SOURCE
-Wp,-MD,.block-qcow2.o.d -c -o block-qcow2.o block-qcow2.c

cc1: warnings being treated as errors
block-qcow2.c: In function Œqcow2_create¹:
block-qcow2.c:1312: warning: ignoring return value of Œwrite¹, declared with attribute warn_unused_result
block-qcow2.c:1314: warning: ignoring return value of Œwrite¹, declared with attribute warn_unused_result
block-qcow2.c:1319: warning: ignoring return value of Œwrite¹, declared with attribute warn_unused_result
block-qcow2.c:1322: warning: ignoring return value of Œwrite¹, declared with attribute warn_unused_result
block-qcow2.c:1325: warning: ignoring return value of Œwrite¹, declared with attribute warn_unused_result
block-qcow2.c: In function Œqcow_write_compressed¹:
block-qcow2.c:1354: warning: ignoring return value of Œftruncate¹, declared with attribute warn_unused_result
.....
.....

Thanks,
Maturi

_______________________________________________
Xen-devel mailing list
[email protected]
http://lists.xensource.com/xen-devel

#2 lore raw
Quoted 3 lines

Properly speaking we should check those write() return values anyway. If we don't want to deviate from the qemu original sources that much then we could special-case CFLAGS to remove -Wno-unused for just that file.

Shouldn't -Wno-unused disable the check rather than enabling it? Removing it causes even more warnings. Maturi's compiler just seems to ignore that switch.

I could compile the code with both gcc 4.1.2 and 4.3.0 without any such problems.

Kevin

#3 lore raw

PATCH 1/1 blktap: Fix unused warnings in block-qcow2.c

Quoted 3 lines · Kevin Wolf

Shouldn't -Wno-unused disable the check rather than enabling it? Removing it causes even more warnings. Maturi's compiler just seems to ignore that switch.

I see, it's the -D_FORTIFY_SOURCE=2 which causes the warnings. This is not enabled in the standard xen-unstable Makefiles, so we could not reproduce it.

The fix is trivial: These warning are all in functions which are used only by qemu, but not by blktap. Just drop them.

Kevin

tools/blktap/drivers/block-qcow2.c +0 -161
diff -r 863563b3e029 tools/blktap/drivers/block-qcow2.c
@@ -1241,167 +1241,6 @@ static void create_refcount_update(QCowC
 		refcount++;
 		*p = cpu_to_be16(refcount);
 	}
-}
-
-static int qcow2_create(const char *filename, int64_t total_size,
-		const char *backing_file, int flags)
-{
-	int fd, header_size, backing_filename_len, l1_size, i, shift, l2_bits;
-	QCowHeader header;
-	uint64_t tmp, offset;
-	QCowCreateState s1, *s = &s1;
-
-	memset(s, 0, sizeof(*s));
-
-	fd = open(filename, O_WRONLY | O_CREAT | O_TRUNC | O_BINARY, 0644);
-	if (fd < 0)
-		return -1;
-	memset(&header, 0, sizeof(header));
-	header.magic = cpu_to_be32(QCOW_MAGIC);
-	header.version = cpu_to_be32(QCOW_VERSION);
-	header.size = cpu_to_be64(total_size * 512);
-	header_size = sizeof(header);
-	backing_filename_len = 0;
-	if (backing_file) {
-		header.backing_file_offset = cpu_to_be64(header_size);
-		backing_filename_len = strlen(backing_file);
-		header.backing_file_size = cpu_to_be32(backing_filename_len);
-		header_size += backing_filename_len;
-	}
-	s->cluster_bits = 12;  /* 4 KB clusters */
-	s->cluster_size = 1 << s->cluster_bits;
-	header.cluster_bits = cpu_to_be32(s->cluster_bits);
-	header_size = (header_size + 7) & ~7;
-	if (flags & BLOCK_FLAG_ENCRYPT) {
-		header.crypt_method = cpu_to_be32(QCOW_CRYPT_AES);
-	} else {
-		header.crypt_method = cpu_to_be32(QCOW_CRYPT_NONE);
-	}
-	l2_bits = s->cluster_bits - 3;
-	shift = s->cluster_bits + l2_bits;
-	l1_size = (((total_size * 512) + (1LL << shift) - 1) >> shift);
-	offset = align_offset(header_size, s->cluster_size);
-	s->l1_table_offset = offset;
-	header.l1_table_offset = cpu_to_be64(s->l1_table_offset);
-	header.l1_size = cpu_to_be32(l1_size);
-	offset += align_offset(l1_size * sizeof(uint64_t), s->cluster_size);
-
-	s->refcount_table = qemu_mallocz(s->cluster_size);
-	if (!s->refcount_table)
-		goto fail;
-	s->refcount_block = qemu_mallocz(s->cluster_size);
-	if (!s->refcount_block)
-		goto fail;
-
-	s->refcount_table_offset = offset;
-	header.refcount_table_offset = cpu_to_be64(offset);
-	header.refcount_table_clusters = cpu_to_be32(1);
-	offset += s->cluster_size;
-
-	s->refcount_table[0] = cpu_to_be64(offset);
-	s->refcount_block_offset = offset;
-	offset += s->cluster_size;
-
-	/* update refcounts */
-	create_refcount_update(s, 0, header_size);
-	create_refcount_update(s, s->l1_table_offset, l1_size * sizeof(uint64_t));
-	create_refcount_update(s, s->refcount_table_offset, s->cluster_size);
-	create_refcount_update(s, s->refcount_block_offset, s->cluster_size);
-
-	/* write all the data */
-	write(fd, &header, sizeof(header));
-	if (backing_file) {
-		write(fd, backing_file, backing_filename_len);
-	}
-	lseek(fd, s->l1_table_offset, SEEK_SET);
-	tmp = 0;
-	for(i = 0;i < l1_size; i++) {
-		write(fd, &tmp, sizeof(tmp));
-	}
-	lseek(fd, s->refcount_table_offset, SEEK_SET);
-	write(fd, s->refcount_table, s->cluster_size);
-
-	lseek(fd, s->refcount_block_offset, SEEK_SET);
-	write(fd, s->refcount_block, s->cluster_size);
-
-	qemu_free(s->refcount_table);
-	qemu_free(s->refcount_block);
-	close(fd);
-	return 0;
-fail:
-	qemu_free(s->refcount_table);
-	qemu_free(s->refcount_block);
-	close(fd);
-	return -ENOMEM;
-}
-
-/* XXX: put compressed sectors first, then all the cluster aligned
-   tables to avoid losing bytes in alignment */
-static int qcow_write_compressed(struct disk_driver *bs, int64_t sector_num,
-		const uint8_t *buf, int nb_sectors)
-{
-	BDRVQcowState *s = bs->private;
-	z_stream strm;
-	int ret, out_len;
-	uint8_t *out_buf;
-	uint64_t cluster_offset;
-
-	if (nb_sectors == 0) {
-		/* align end of file to a sector boundary to ease reading with
-		   sector based I/Os */
-		cluster_offset = 512 * s->total_sectors;
-		cluster_offset = (cluster_offset + 511) & ~511;
-		ftruncate(s->fd, cluster_offset);
-		return 0;
-	}
-
-	if (nb_sectors != s->cluster_sectors)
-		return -EINVAL;
-
-	out_buf = qemu_malloc(s->cluster_size + (s->cluster_size / 1000) + 128);
-	if (!out_buf)
-		return -ENOMEM;
-
-	/* best compression, small window, no zlib header */
-	memset(&strm, 0, sizeof(strm));
-	ret = deflateInit2(&strm, Z_DEFAULT_COMPRESSION,
-			Z_DEFLATED, -12,
-			9, Z_DEFAULT_STRATEGY);
-	if (ret != 0) {
-		qemu_free(out_buf);
-		return -1;
-	}
-
-	strm.avail_in = s->cluster_size;
-	strm.next_in = (uint8_t *)buf;
-	strm.avail_out = s->cluster_size;
-	strm.next_out = out_buf;
-
-	ret = deflate(&strm, Z_FINISH);
-	if (ret != Z_STREAM_END && ret != Z_OK) {
-		qemu_free(out_buf);
-		deflateEnd(&strm);
-		return -1;
-	}
-	out_len = strm.next_out - out_buf;
-
-	deflateEnd(&strm);
-
-	if (ret != Z_STREAM_END || out_len >= s->cluster_size) {
-		/* could not compress: write normal cluster */
-		qcow_write(bs, sector_num, buf, s->cluster_sectors);
-	} else {
-		cluster_offset = get_cluster_offset(bs, sector_num << 9, 2,
-											out_len, 0, 0);
-		cluster_offset &= s->cluster_offset_mask;
-		if (bdrv_pwrite(s->fd, cluster_offset, out_buf, out_len) != out_len) {
-			qemu_free(out_buf);
-			return -1;
-		}
-	}
-
-	qemu_free(out_buf);
-	return 0;
 }
 
 static int qcow_submit(struct disk_driver *bs)

_______________________________________________
Xen-devel mailing list
[email protected]
http://lists.xensource.com/xen-devel

#4 lore raw
In reply to Kevin Wolf
Quoted 10 lines · Kevin Wolf
Quoted 3 lines

Properly speaking we should check those write() return values anyway. If we don't want to deviate from the qemu original sources that much then we could special-case CFLAGS to remove -Wno-unused for just that file.

Shouldn't -Wno-unused disable the check rather than enabling it? Removing it causes even more warnings. Maturi's compiler just seems to ignore that switch.

I could compile the code with both gcc 4.1.2 and 4.3.0 without any such problems.

Oh yes, silly me.

-- Keir

#6 lore raw
Quoted 1 line · Srinivas Maturi

We are using gcc 4.1.2.

Ok. As I wrote the problem is not with the compiler version but with your compiler switches, namely -D_FORTIFY_SOURCE=2. Keir has commited my patch already, so please try the latest xen-unstable.

Kevin

#7 lore raw
Quoted 11 lines · Kevin Wolf
Quoted 2 lines · Srinivas Maturi

We are using gcc 4.1.2.

Ok. As I wrote the problem is not with the compiler version but with your compiler switches, namely -D_FORTIFY_SOURCE=2. Keir has commited my patch already, so please try the latest xen-unstable.

The build succeeded with changeset 17152. Thanks for the patch.

Thanks,
Maturi