| Message ID | 1475064333-22504-10-git-send-email-hegdevasant@linux.vnet.ibm.com |
|---|---|
| State | Superseded |
| Headers | show |
On Wed, Sep 28, 2016 at 10:05 PM, Vasant Hegde <hegdevasant@linux.vnet.ibm.com> wrote: > Hostboot reserves some memory and passes via HDAT. This patch > adds code to parse node reservation details and marks it as > REGION_HW_RESERVED. Later mem_region_add_dt_reserved() populates > DT entry. > > Signed-off-by: Vasant Hegde <hegdevasant@linux.vnet.ibm.com> > --- > hdata/memory.c | 31 +++++++++++++++++++++++++++++++ > 1 file changed, 31 insertions(+) > > diff --git a/hdata/memory.c b/hdata/memory.c > index a8b9955..bb776fd 100644 > --- a/hdata/memory.c > +++ b/hdata/memory.c > @@ -19,6 +19,7 @@ > #include <vpd.h> > #include <ccan/str/str.h> > #include <libfdt/libfdt.h> > +#include <mem_region.h> > #include <types.h> > > #include "spira.h" > @@ -383,6 +384,34 @@ static void get_msareas(struct dt_node *root, > } > } > > +static void get_hb_reserved_mem(struct HDIF_common_hdr *ms_vpd) > +{ > + int count, i; > + const struct msvpd_hb_reserved_mem *hb_resv_mem; > + > + if (proc_gen < proc_gen_p9) > + return; > + > + count = HDIF_get_iarray_size(ms_vpd, MSVPD_IDATA_HB_RESERVED_MEM); > + if (count <= 0) { > + prerror("MS VPD: No hostboot reserved memory found\n"); > + return; > + } > + > + for (i = 0; i < count; i++) { > + hb_resv_mem = HDIF_get_iarray_item(ms_vpd, > + MSVPD_IDATA_HB_RESERVED_MEM, > + i, NULL); > + if (!CHECK_SPPTR(hb_resv_mem)) > + continue; > + > +#if !defined(TEST) > + mem_reserve_hw(hb_resv_mem->label, hb_resv_mem->start_addr, I'm not sure it's safe to use hb_res_mem->label directly as a string. The spec doesn't say it needs to be NUL terminated and expects you to parse it based on the length field. > + (hb_resv_mem->end_addr - hb_resv_mem->start_addr)); Most of the existing parsing code uses be*_to_cpu() rather than reading the values directly. That said I don't think we have any plans to make skiboot run little endian so I'm not sure I care... > +#endif > + } > +} > + > static bool __memory_parse(struct dt_node *root) > { > struct HDIF_common_hdr *ms_vpd; > @@ -431,6 +460,8 @@ static bool __memory_parse(struct dt_node *root) > > get_msareas(root, ms_vpd); > > + get_hb_reserved_mem(ms_vpd); > + > prlog(PR_INFO, "MS VPD: Total MB of RAM: 0x%llx\n", > (long long)be64_to_cpu(tcms->total_in_mb)); > > -- > 2.5.5 > > _______________________________________________ > Skiboot mailing list > Skiboot@lists.ozlabs.org > https://lists.ozlabs.org/listinfo/skiboot
On 09/29/2016 07:58 AM, Oliver O'Halloran wrote: > On Wed, Sep 28, 2016 at 10:05 PM, Vasant Hegde > <hegdevasant@linux.vnet.ibm.com> wrote: .../... >> + >> +#if !defined(TEST) >> + mem_reserve_hw(hb_resv_mem->label, hb_resv_mem->start_addr, > > I'm not sure it's safe to use hb_res_mem->label directly as a string. > The spec doesn't say it needs to be NUL terminated and expects you to > parse it based on the length field. Ah! I assumed it will terminate with NULL :-) I agree that its not safe to assume specially on someone else is passing data :-) Will fix in V2. > >> + (hb_resv_mem->end_addr - hb_resv_mem->start_addr)); > > Most of the existing parsing code uses be*_to_cpu() rather than > reading the values directly. That said I don't think we have any plans > to make skiboot run little endian so I'm not sure I care... Its always BE. So many a time we forget to use converters .. But its good to use. will fix in v2. -Vasant
diff --git a/hdata/memory.c b/hdata/memory.c index a8b9955..bb776fd 100644 --- a/hdata/memory.c +++ b/hdata/memory.c @@ -19,6 +19,7 @@ #include <vpd.h> #include <ccan/str/str.h> #include <libfdt/libfdt.h> +#include <mem_region.h> #include <types.h> #include "spira.h" @@ -383,6 +384,34 @@ static void get_msareas(struct dt_node *root, } } +static void get_hb_reserved_mem(struct HDIF_common_hdr *ms_vpd) +{ + int count, i; + const struct msvpd_hb_reserved_mem *hb_resv_mem; + + if (proc_gen < proc_gen_p9) + return; + + count = HDIF_get_iarray_size(ms_vpd, MSVPD_IDATA_HB_RESERVED_MEM); + if (count <= 0) { + prerror("MS VPD: No hostboot reserved memory found\n"); + return; + } + + for (i = 0; i < count; i++) { + hb_resv_mem = HDIF_get_iarray_item(ms_vpd, + MSVPD_IDATA_HB_RESERVED_MEM, + i, NULL); + if (!CHECK_SPPTR(hb_resv_mem)) + continue; + +#if !defined(TEST) + mem_reserve_hw(hb_resv_mem->label, hb_resv_mem->start_addr, + (hb_resv_mem->end_addr - hb_resv_mem->start_addr)); +#endif + } +} + static bool __memory_parse(struct dt_node *root) { struct HDIF_common_hdr *ms_vpd; @@ -431,6 +460,8 @@ static bool __memory_parse(struct dt_node *root) get_msareas(root, ms_vpd); + get_hb_reserved_mem(ms_vpd); + prlog(PR_INFO, "MS VPD: Total MB of RAM: 0x%llx\n", (long long)be64_to_cpu(tcms->total_in_mb));
Hostboot reserves some memory and passes via HDAT. This patch adds code to parse node reservation details and marks it as REGION_HW_RESERVED. Later mem_region_add_dt_reserved() populates DT entry. Signed-off-by: Vasant Hegde <hegdevasant@linux.vnet.ibm.com> --- hdata/memory.c | 31 +++++++++++++++++++++++++++++++ 1 file changed, 31 insertions(+)