Opened 15 years ago
Closed 8 years ago
#5146 closed bug (fixed)
get_nth_pci_info missing
Reported by: | andrewbachmann | Owned by: | axeld |
---|---|---|---|
Priority: | low | Milestone: | R1 |
Component: | System/libroot.so | Version: | R1/Development |
Keywords: | get_nth_pci_info | Cc: | |
Blocked By: | Blocking: | ||
Platform: | All |
Description (last modified by )
I was trying to run BeRometer to do some performance testing and it complains about this symbol missing. The symbol is in libroot.so on BeOS R 5.
Attachments (5)
Change History (19)
follow-up: 2 comment:1 by , 15 years ago
comment:2 by , 15 years ago
Replying to axeld:
But it's also private API. Nevertheless, a nice API to enumerate the hardware (something more/better than the info that is in the device tree) would be appreciated.
Yeah, it's private API. Some other programs are also using it. Should we add a stub for the method/symbol and get around to flushing it out later when doing the stack of nice-to-haves?
by , 13 years ago
Attachment: | get_pci_nth.patch added |
---|
BeRometer works well now, but it needs a libmail from BeOS R5 becose it using some undocumented API from it, patch for haiku`s libmail is work-in-progress
follow-up: 4 comment:3 by , 13 years ago
patch: | 0 → 1 |
---|
follow-up: 6 comment:4 by , 13 years ago
Replying to BeSman:
BeRometer works well now, but it needs a libmail from BeOS R5 becose it using some undocumented API from it, patch for haiku`s libmail is work-in-progress
long get_nth_pci_info(long index,pci_info *info) { pci_module_info *mod_info; if(mod_info->get_nth_pci_info(index,info)) return B_NO_ERROR; else return B_ERROR; }
Please explain what are you trying to do here. Thanks.
BTW, read the Coding Guidelines too. ;-)
comment:5 by , 13 years ago
Im provided a BeOS R5 compatible callback for get_nth_pci_info() function which we have in pci bus manager but not in libroot.so
comment:6 by , 13 years ago
mod_info is unset, your code will most probably crash at pointer dereference on next line. You clearly never test this code ;-)
comment:7 by , 13 years ago
patch: | 1 → 0 |
---|
comment:8 by , 13 years ago
Description: | modified (diff) |
---|
comment:9 by , 13 years ago
patch: | 0 → 1 |
---|
follow-ups: 11 12 comment:10 by , 13 years ago
Version: | R1/alpha1 → R1/Development |
---|
by , 13 years ago
Attachment: | 0001-Added-stub-for-get_nth_pci_info.-BeRometer-works-now.patch added |
---|
Added stub for get_nth_pci_info. BeRometer works now.
comment:11 by , 13 years ago
by , 13 years ago
Attachment: | BeRometer.results_without_whetstone added |
---|
by , 13 years ago
Attachment: | BeRometer_bt.JPG added |
---|
comment:12 by , 13 years ago
by , 9 years ago
Attachment: | BeRometer.zip added |
---|
BeRometer binary with a patched libroot.so. Whetstone test deadlocks Haiku.
comment:14 by , 8 years ago
Resolution: | → fixed |
---|---|
Status: | new → closed |
Patch applied in hrev50554. BeRoMeter now works. I didn't retry the whetstone test. I also noticed that the BeRoMeter installer complains that I have only -94MB of RAM. I guess it overflows somewhere when there are more than 2GB on the system, but I could test diver's binary.
But it's also private API. Nevertheless, a nice API to enumerate the hardware (something more/better than the info that is in the device tree) would be appreciated.