* Koen Kooi <[EMAIL PROTECTED]> [081004 12:39]: > > Op 2 okt 2008, om 11:29 heeft Kevin Hilman het volgende geschreven: > >> Mans Rullgard <[EMAIL PROTECTED]> writes: >> >>> This increases VMALLOC_END to 0x18000000, making room for 256MB >>> RAM with the default 128MB vmalloc region. >>> >>> Signed-off-by: Mans Rullgard <[EMAIL PROTECTED]> >>> --- >>> arch/arm/plat-omap/include/mach/vmalloc.h | 2 +- >>> 1 files changed, 1 insertions(+), 1 deletions(-) >>> >>> diff --git a/arch/arm/plat-omap/include/mach/vmalloc.h b/arch/arm/ >>> plat-omap/include/mach/vmalloc.h >>> index d8515cb..b97dfaf 100644 >>> --- a/arch/arm/plat-omap/include/mach/vmalloc.h >>> +++ b/arch/arm/plat-omap/include/mach/vmalloc.h >>> @@ -17,5 +17,5 @@ >>> * along with this program; if not, write to the Free Software >>> * Foundation, Inc., 59 Temple Place, Suite 330, Boston, MA >>> 02111-1307 USA >>> */ >>> -#define VMALLOC_END (PAGE_OFFSET + 0x17000000) >>> +#define VMALLOC_END (PAGE_OFFSET + 0x18000000) >> >> Acked-by: Kevin Hilman <[EMAIL PROTECTED]> >> >> I have a similar patch locally, but the problem I currently have with >> it is that there is no longer any hole between vmalloc space and the >> beginning of IO space (the first virtual mappings start at >> 0xd8000000.) >> >> It's a bit unsafe, but IMO it's is probably OK for the short term. >> Longer term I think the virtual address space of the OMAP kernels >> needs to be reworked. It currenly just maps directly onto the >> physical address space, which makes the IO_ADDRESS() conversion macros >> simple, but leaves lots of wasted space in the virtual address space.
Seems to be also safe for omap1, so I've pushed it. > I was told that omap3 support 512MB of ram, so is it possible to account > for that as well, or can we revisit that when the actual LP-DDR become > available? Well, to make room for a long term solution let's do the following: - Use OMAP1_IO_ADDRESS and OMAP2_IO_ADDRES and get rid of OMAP_IO_ADDRESS. This allows compiling in more code into a single binary - Remove the remaining io_v2p() and any XXX_IO_ADDRESS() in the drivers. Drivers should be now using ioremap() as noted earlier - Then we can further split OMAP2_IO_ADDRESS() into L3_IO_ADDRESS() and L4_IO_ADDRESS(), which allows us to move L4_XXXX_VIRT out of the way Anybody got better ideas? Regards, Tony -- To unsubscribe from this list: send the line "unsubscribe linux-omap" in the body of a message to [EMAIL PROTECTED] More majordomo info at http://vger.kernel.org/majordomo-info.html
