My platform is the Overo Fire.
Solution further down.
The sd slot and wireless/bluetooth are both connected to the mmc
interface at:
/sys/class/mmc_host/mmc0
and
/sys/class/mmc_host/mmc1
the sd card located at mmc0 bootstraps just fine. When VOLD attempts
to also bootstrap mmc1, segfault occurs:
bootstrap_card(/sys/class/mmc_host/mmc1/mmc1:0001):
I/DEBUG ( 888): *** *** *** *** *** *** *** *** *** *** *** *** ***
*** *** ***
I/DEBUG ( 888): Build fingerprint: 'generic/generic/generic/:1.5/
CUPCAKE/eng.demo.20090716.170157:eng/test-keys'
I/DEBUG ( 888): pid: 887, tid: 887 >>> /system/bin/vold <<<
I/DEBUG ( 888): signal 11 (SIGSEGV), fault addr 00000000
I/DEBUG ( 888): r0 00000000 r1 00020000 r2 00000000 r3 80808080
I/DEBUG ( 888): r4 00000000 r5 00000000 r6 beb40480 r7 000121cc
I/DEBUG ( 888): r8 00000000 r9 00000000 10 00000000 fp 00000000
I/DEBUG ( 888): ip 00000000 sp beb40458 lr 0000a813 pc
afe0e188 cpsr 40000010
I/DEBUG ( 888): #00 pc 0000e188 /system/lib/libc.so
I/DEBUG ( 888): #01 pc 0000a810 /system/bin/vold
I/DEBUG ( 888): #02 pc 0000a9be /system/bin/vold
I/DEBUG ( 888): #03 pc 0000aa80 /system/bin/vold
I/DEBUG ( 888): #04 pc 000097c4 /system/bin/vold
I/DEBUG ( 888): #05 pc 0001fd6a /system/lib/libc.so
I/DEBUG ( 888): #06 pc 0000bcb2 /system/lib/libc.so
I/DEBUG ( 888): #07 pc b000157e /system/bin/linker
I/DEBUG ( 888): stack:
I/DEBUG ( 888): beb40418 000000dc
I/DEBUG ( 888): beb4041c 00002bb0
this is the content of my vold.conf:
volume_sdcard {
## This is the direct uevent device path to the SD slot on the
device
media_path /devices/platform/mmci-omap-hs.0/mmc_host/mmc0
media_type mmc
mount_point /sdcard
ums_path /devices/platform/usb_mass_storage/lun0
}
To fix this, I kind of went around the issue all together since I
didn't want to spend forever pinpointing the line that was causing the
bootstrap process to segfault on the sdio wireless chip.
First change is mandatory. Second change is not needed.
--- a/system/core/vold/mmc.c 2009-07-02 16:15:08.000000000 -0400
+++ b/system/core/vold/mmc.c 2009-07-17 10:35:02.000000000 -0400
@@ -54,8 +54,14 @@
continue;
sprintf(tmp, "%s/%s", SYSFS_CLASS_MMC_PATH, de->d_name);
- if (mmc_bootstrap_controller(tmp))
+ if (mmc_bootstrap_controller(tmp)) {
LOG_ERROR("Error bootstrapping controller '%s' (%m)",
tmp);
+ }
+ else {
+ //a controller has already been bootstrapped, lets stop
trying to get more
+ //TODO: hack fix to get around segfault on boostrapping
sdio card connected to mmc bus
+ break;
+ }
}
closedir(d);
@@ -85,7 +91,8 @@
if ((!strcmp(de->d_name, "uevent")) ||
(!strcmp(de->d_name, "subsystem")) ||
(!strcmp(de->d_name, "device")) ||
- (!strcmp(de->d_name, "power"))) {
+ (!strcmp(de->d_name, "power")) ||
+ (!strcmp(de->d_name, "slot_name"))) {
continue;
}
I hope this saves people some time when trying to get VOLD to work
certain systems. I don't know why VOLD is even attempting to bootstrap
mmc1 when only mmc0 is clearly defined in the vold.conf file. Possibly
a bug?
--~--~---------~--~----~------------~-------~--~----~
unsubscribe: android-porting+unsubscr...@googlegroups.com
website: http://groups.google.com/group/android-porting
-~----------~----~----~----~------~----~------~--~---
