usb: Fix clang build
PATCH 1/1 usb: Fix clang build
../hw/usb/dev-uas.c:157:31: error: field 'status' with variable sized type 'uas_iu' not at the end of a struct or class is a GNU extension [-Werror,-Wgnu-variable-sized-type-not-at-end]
uas_iu status;
^Fix this by specifying a size for the add_cdb member; and at present, the code does not actually use that field other than for the size chosen for the packed uas_iu_command struct, and the choice of one byte does not change the size of the uas_iu union.
---
I'm not sure why none of our CI tools pick up this particular clang build failure; I hit it on Fedora 33 when configuring to build the entire tree with clang.
hw/usb/dev-uas.c | 2 +- 1 file changed, 1 insertion(+), 1 deletion(-)
hw/usb/dev-uas.c +1 -1
@@ -70,7 +70,7 @@ typedef struct {uint8_t reserved_2;uint64_t lun;uint8_t cdb[16 ]; -uint8_t add_cdb[]; +uint8_t add_cdb[1 ]; }QEMU_PACKED uas_iu_command; typedef struct {
Signature · 1 line
2.30.0
Hi Eric,
Quoted 8 lines · Eric Blake
../hw/usb/dev-uas.c:157:31: error: field 'status' with variable sized type 'uas_iu' not at the end of a struct or class is a GNU extension [-Werror,-Wgnu-variable-sized-type-not-at-end]uas_iu status; ^Fix this by specifying a size for the
add_cdbmember; and at present, the code does not actually use that field other than for the size chosen for the packeduas_iu_commandstruct, and the choice of one byte does not change the size of theuas_iuunion.
I sent a maybe safer approach (from the bus PoV):
https://www.mail-archive.com/[email protected]/msg79192.html
Do you mind reviewing it?
Quoted 7 lines
Signed-off-by Eric Blake <[email protected]>---
I'm not sure why none of our CI tools pick up this particular clang build failure; I hit it on Fedora 33 when configuring to build the entire tree with clang.
Same issue after upgrading to f33. I sent a patch to bump our CI:
https://www.mail-archive.com/[email protected]/msg774117.html
To track Fedora releases I was thinking about a gitlab job checking if we are using the latest, else failing; smth as:
$ curl https://getfedora.org/ | grep -q 'Fedora 33 released'
Quoted 12 lines · Philippe Mathieu-Daudé
Hi Eric,
Quoted 8 lines · Eric Blake
../hw/usb/dev-uas.c:157:31: error: field 'status' with variable sized type 'uas_iu' not at the end of a struct or class is a GNU extension [-Werror,-Wgnu-variable-sized-type-not-at-end]uas_iu status; ^Fix this by specifying a size for the
add_cdbmember; and at present, the code does not actually use that field other than for the size chosen for the packeduas_iu_commandstruct, and the choice of one byte does not change the size of theuas_iuunion.
Quoted 12 lines
I sent a maybe safer approach (from the bus PoV):
https://www.mail-archive.com/[email protected]/msg79192.htmlDo you mind reviewing it?
Quoted 7 lines
Signed-off-by Eric Blake <[email protected]>---
I'm not sure why none of our CI tools pick up this particular clang build failure; I hit it on Fedora 33 when configuring to build the entire tree with clang.
BTW first report is from 28 Sep 2020 (Ed):
https://www.mail-archive.com/[email protected]/msg745525.html
Then on 23 Oct 2020 (Daniele):
https://www.mail-archive.com/[email protected]/msg753674.html
Then on 10 Nov 2020 (Han):
https://www.mail-archive.com/[email protected]/msg759108.html
Quoted 8 lines
Same issue after upgrading to f33. I sent a patch to bump our CI:
https://www.mail-archive.com/[email protected]/msg774117.htmlTo track Fedora releases I was thinking about a gitlab job checking if we are using the latest, else failing; smth as:
$ curl https://getfedora.org/ | grep -q 'Fedora 33 released'
Quoted 25 lines · Philippe Mathieu-Daudé
Hi Eric,
Quoted 8 lines · Eric Blake
../hw/usb/dev-uas.c:157:31: error: field 'status' with variable sized type 'uas_iu' not at the end of a struct or class is a GNU extension [-Werror,-Wgnu-variable-sized-type-not-at-end]uas_iu status; ^Fix this by specifying a size for the
add_cdbmember; and at present, the code does not actually use that field other than for the size chosen for the packeduas_iu_commandstruct, and the choice of one byte does not change the size of theuas_iuunion.I sent a maybe safer approach (from the bus PoV):
https://www.mail-archive.com/[email protected]/msg79192.htmlDo you mind reviewing it?
Quoted 7 lines
Signed-off-by Eric Blake <[email protected]>---
I'm not sure why none of our CI tools pick up this particular clang build failure; I hit it on Fedora 33 when configuring to build the entire tree with clang.
Quoted 7 lines
Same issue after upgrading to f33. I sent a patch to bump our CI:
https://www.mail-archive.com/[email protected]/msg774117.htmlTo track Fedora releases I was thinking about a gitlab job checking if we are using the latest, else failing; smth as:
$ curl https://getfedora.org/ | grep -q 'Fedora 33 released'
We'd be better off just adding a job that targets Fedora rawhide IMHO so we get immediate notice that a toolchain update is going to hurt us, instead of waiting until the next Fedora is already released to find the problem.
Regards,
Daniel
Signature · 3 lines
|: https://berrange.com -o- https://www.flickr.com/photos/dberrange :| |: https://libvirt.org -o- https://fstop138.berrange.com :| |: https://entangle-photo.org -o- https://www.instagram.com/dberrange :|