Implement paging on the pci arbiter
Hi, I'm doing a research on how to allow arbiter clients to map bar regions into their spaces. It seems to me the approach should be: 1.- Make the arbiter implement paging 2.- Expect clients to mmap() the region files the arbiter generates Sounds that correct for you? I'm a bit confused with libpager, the reference manual seems to be obsolete: https://www.gnu.org/software/hurd/doc/hurd_5.html#SEC32 From this document I assumed I should create a pager for each region file, call page_demuxer from the arbiter's demuxer and implement callback functions. But it seems pager_demuxer() is not defined anymore and servers use pager_start_workers() which I'm not sure what's for. Is there more recent documentation about libpager? signature.asc Description: OpenPGP digital signature
Re: PCI arbiter crash on last qemu image
Hi,
On 23/8/20 8:47 pm, Joan Lledó wrote:
> http://git.savannah.gnu.org/cgit/hurd/hurd.git/log/?h=jlledom-pciaccess-map
Thanks for doing this, I tried it locally and fixed two bugs in my libpciaccess
patches:
diff --git a/src/x86_pci.c b/src/x86_pci.c
index 1614729..1e70f35 100644
--- a/src/x86_pci.c
+++ b/src/x86_pci.c
@@ -275,6 +275,7 @@ map_dev_mem(void **dest, size_t mem_offset, size_t
mem_size, int write)
return errno;
}
+close(memfd);
return 0;
#endif
}
@@ -505,7 +506,7 @@ pci_nfuncs(struct pci_device *dev, uint8_t *nfuncs)
static error_t
pci_device_x86_read_rom(struct pci_device *dev, void *buffer)
{
-void *bios;
+void *bios = NULL;
struct pci_device_private *d = (struct pci_device_private *)dev;
int err;
> Also in map_dev_mem(), it seems to be some problem when mapping the rom.
> I tried to read the rom with hexdump:
>
> hexdump -Cn 16 /servers/bus/pci//00/03/0/rom
This command now works (every time).
> In pci_device_x86_read_rom() the memory is mapped and unmapped for each
> read. I wonder if it's correct to unmap with munmap() something mapped
> with vm_map()
I think it's okay to munmap memory mapped with vm_map
but I forgot to initialise the rom memory pointer to NULL.
See http://git.zammit.org/libpciaccess.git/log/?h=rumpdisk-upstream again for
updated branch.
If you think everything is okay with this, I will squash the last patch and
submit patches upstream.
Thanks,
Damien
doc libpager
Hi Joan, I can not reply in your thread "Implement paging on the pci arbiter" because i suscribed right now. hope you this links! This doc talk about libpager. https://www.freesoft.org/software/hurd/building.pdf (https://www.freesoft.org/software/hurd/building.pdf) https://github.com/BrentBaccala/hurd (https://github.com/BrentBaccala/hurd)
Re: Implement paging on the pci arbiter
Hello, Joan Lledó, le mer. 26 août 2020 10:24:49 +0200, a ecrit: > I'm doing a research on how to allow arbiter clients to map bar regions > into their spaces. It seems to me the approach should be: > > 1.- Make the arbiter implement paging > 2.- Expect clients to mmap() the region files the arbiter generates > > Sounds that correct for you? > > I'm a bit confused with libpager, I'm unsure if libpager will be useful actually, since all you need is to pass on a memory object clamped to the target physical memory. See gnumach's support for proxy memory object, which possibly is just enough. Samuel
Re: doc libpager
Hi, thanks for your help, I'll take a look :) El 26/8/20 a les 16:23, gfle...@disroot.org ha escrit: > > Hi Joan, > > I can not reply in your thread "Implement paging on the pci arbiter" > because i suscribed right now. hope you this links! > > This doc talk about libpager. > > https://www.freesoft.org/software/hurd/building.pdf > https://github.com/BrentBaccala/hurd >
Re: PCI arbiter crash on last qemu image
Hi, El 26/8/20 a les 11:13, Damien Zammit ha escrit: > If you think everything is okay with this, I will squash the last patch and > submit patches upstream. Yes it's OK for me
Re: Implement paging on the pci arbiter
Joan Lledó, le mer. 26 août 2020 10:24:49 +0200, a ecrit: > From this document I assumed I should create a pager for each region > file, call page_demuxer from the arbiter's demuxer and implement > callback functions. But it seems pager_demuxer() is not defined anymore ? libpager/demuxer.c:pager_demuxer (struct pager_requests *requests, > and servers use pager_start_workers() which I'm not sure what's for. It's just a way to manage several threads to handle RPCs. See how it creates a thread than calls service_paging_requests that runs pager_demuxer. Samuel
