[Bluez-users] sdpd crashing (in uclinux)

[Bluez-users] sdpd crashing (in uclinux)

[Jorge A. González]

Hello

I have already "ported" the bluez-utils to uclinux (over an ARM7DMI)
and I realized than sdpd crash whenever other device tries to browse
its services). I have printfed the code and followed it until I found
out that it is the command sdp_put_unligned which is having a
segmentation fault.
I have seen that sdp_put_unligned is defined as a macro depending of
the definition or not of 386 architecture. I will trie to play with
the defines and to port hcidump in order to have more information but
I would like to ask the list about this problem in the case I am not the
first one.

(I think my system is little endian, but I will also trie to be sure of
it today)

Thanks for your help!

Jorge



-------------------------------------------------------
This SF.Net email is sponsored by: SourceForge.net Broadband
Sign-up now for SourceForge Broadband and get the fastest
6.0/768 connection for only $19.95/mo for the first 3 months!
http://ads.osdn.com/?ad_id=2562&alloc_id=6184&op=click
_______________________________________________
Bluez-users mailing list
Bluez-users@lists.sourceforge.net
https://lists.sourceforge.net/lists/listinfo/bluez-users

Re: [Bluez-users] sdpd crashing (in uclinux)

[Marcel Holtmann]

Hi Jorge,

> I have already "ported" the bluez-utils to uclinux (over an ARM7DMI)
> and I realized than sdpd crash whenever other device tries to browse
> its services). I have printfed the code and followed it until I found
> out that it is the command sdp_put_unligned which is having a
> segmentation fault.
> I have seen that sdp_put_unligned is defined as a macro depending of
> the definition or not of 386 architecture. I will trie to play with
> the defines and to port hcidump in order to have more information but
> I would like to ask the list about this problem in the case I am not the
> first one.

please start using bluez-libs-2.7 and bluez-utils-2.7, because the SDP
stuff is now part of the core packages and I stopped maintaining the
bluez-sdp package.

The new Bluetooth library contains a bt_put_unaligned macro for general
use and I used a slight different method for non i386 platforms.

Regards

Marcel




-------------------------------------------------------
This SF.Net email is sponsored by: SourceForge.net Broadband
Sign-up now for SourceForge Broadband and get the fastest
6.0/768 connection for only $19.95/mo for the first 3 months!
http://ads.osdn.com/?ad_id=2562&alloc_id=6184&op=click
_______________________________________________
Bluez-users mailing list
Bluez-users@lists.sourceforge.net
https://lists.sourceforge.net/lists/listinfo/bluez-users

Re: [Bluez-users] sdpd crashing (in uclinux)

[Marcel Holtmann]

Hi Jorge,

> > please start using bluez-libs-2.7 and bluez-utils-2.7, because the SDP
> > stuff is now part of the core packages and I stopped maintaining the
> > bluez-sdp package.
> > 
> > The new Bluetooth library contains a bt_put_unaligned macro for general
> > use and I used a slight different method for non i386 platforms.
> 
> I am using bluez-libs-2.7 and bluez-utils-2.7. When I refered to sdpd I
> mean the one is located in bluez-utils-2.7/sdpd. Sorry for not saying it
> before.
> 
> The guilty instruction is:
> sdp_put_unaligned(htons(rsp_count), (uint16_t *)pTotalRecordCount);
> 
> in the line 315 of bluez-utils-2.7/sdpd/request.c
> 
> When I reach this line (because other device is trying to browse the
> local services) I get a SIGSEV so sdpd finish.
> 
> The commands I use to reach this state are:
> 
> #hcid -n -f /nfs/hcid.conf &
> #hciattach -n /dev/tts/2 any 115200 flow &
> #hciconfig hci0 up
> #sdpd -n
> 
> And the output of sdpd is:
> 
> ****  A pocket PC is trying to browse our local services
> 
> Got a svc srch req
> Seq type : 53
> Data size : 3
> Data type: 0x19
> No of elements : 1
> Expected count: 20
> Bytes scanned : 5
> Continuation State size : 0
> Checking svcRec : 0x0
> Checking svcRec : 0xb1a038
> Match count: 2
> ***** reached line 315 of request.c
> SIGSEGV
> 
> The instruction sdp_put_unaligned is replaced with the macro
> bt_put_unaligned in the not i386 version and it is  
> "memcpy((ptr), &__tmp, sizeof(*(ptr))); "
> which is crashing. There are other times that this macro is called and
> works normally, so I suppose it is a problem with the pTotalRecordCount
> pointer.
> 
> I have found out than if just after the
> pTotalRecordCount = (short*) pdata;
> 
> I do:
> 
> 	printf("pdata is %p and have %i\n",
> 	       pdata,*pdata);
> 	--this work
> 
> 	printf("pTotalRecordCount is %p and have %i\n",
> 	       pTotalRecordCount,*pTotalRecordCount));
> 	--this give me a SIGSEV!!!!
> 
> but if I write this:
> 	printf("pTotalRecordCount vale %p y contiene %i\n",
> 	       pTotalRecordCount,*((char*) pTotalRecordCount));
> this work!
> 
> This make clear that the problem is with the type of the pointers, but I
> have no idea of how to solve it, because pTotalRecordCount and similar
> ones are casted to uint16_t when they crash.
> 
> Besides these printf the only modification i have made to the code is
> defining the SDP_UNIX_PATH as "/ram/sdp" instead than "/var/run/sdp"
> because /ram is the only place where I have write access.
> 
> My compiler is arm-elf-gcc, the CPU is ARM7DMI and the kernel version is
> 2.4.22.
> 
> I hope all this information help in figuring out the problem

actually I have no idea why this fails. Does it help if you declare
pTotalRecordCount as uint16_t instead of short?

Do the old sdp_put_unaligned code for non i386 platforms from the
bluez-sdp package is working correct?

Regards

Marcel




-------------------------------------------------------
This SF.Net email is sponsored by: Oracle 10g
Get certified on the hottest thing ever to hit the market... Oracle 10g. 
Take an Oracle 10g class now, and we'll give you the exam FREE.
http://ads.osdn.com/?ad_id=3149&alloc_id=8166&op=click
_______________________________________________
Bluez-users mailing list
Bluez-users@lists.sourceforge.net
https://lists.sourceforge.net/lists/listinfo/bluez-users