| Message ID | 26384fad6f9afcba9d3311d1bf73e76f4577225d.1563276298.git.anatoly.burakov@intel.com (mailing list archive) |
|---|---|
| State | Superseded, archived |
| Delegated to: | David Marchand |
| Headers | show |
| Series | eal/freebsd: add support for base virtaddr option | expand |
| Context | Check | Description |
|---|---|---|
| ci/checkpatch | success | coding style OK |
| ci/intel-Performance-Testing | success | Performance Testing PASS |
| ci/mellanox-Performance-Testing | success | Performance Testing PASS |
| ci/Intel-compilation | success | Compilation OK |
16/07/2019 13:25, Anatoly Burakov: > According to our docs, only Linuxapp supports base-virtaddr option. > That is, strictly speaking, not true because most of the things > that are attempting to respect base-virtaddr are in common files, > so FreeBSD already *mostly* supports this option in practice. > > This commit fixes the remaining bits to explicitly support > base-virtaddr option, and moves the arg parsing from EAL to common > options parsing code. Documentation is also updated to reflect > that all platforms now support base-virtaddr. > > Signed-off-by: Anatoly Burakov <anatoly.burakov@intel.com> This patch is deferred to 19.11.
On Tue, Jul 16, 2019 at 1:25 PM Anatoly Burakov <anatoly.burakov@intel.com> wrote: > > According to our docs, only Linuxapp supports base-virtaddr option. > That is, strictly speaking, not true because most of the things > that are attempting to respect base-virtaddr are in common files, > so FreeBSD already *mostly* supports this option in practice. > > This commit fixes the remaining bits to explicitly support > base-virtaddr option, and moves the arg parsing from EAL to common > options parsing code. Documentation is also updated to reflect > that all platforms now support base-virtaddr. > > Signed-off-by: Anatoly Burakov <anatoly.burakov@intel.com> > --- > doc/guides/linux_gsg/eal_args.include.rst | 6 +++ > doc/guides/linux_gsg/linux_eal_parameters.rst | 6 --- > doc/guides/rel_notes/release_19_08.rst | 5 +++ > lib/librte_eal/common/eal_common_options.c | 38 ++++++++++++++++++ > lib/librte_eal/freebsd/eal/eal.c | 13 ++++++- > lib/librte_eal/linux/eal/eal.c | 39 ------------------- > 6 files changed, 60 insertions(+), 47 deletions(-) > > diff --git a/doc/guides/linux_gsg/eal_args.include.rst b/doc/guides/linux_gsg/eal_args.include.rst > index cf421a56e..ed8b0e35b 100644 > --- a/doc/guides/linux_gsg/eal_args.include.rst > +++ b/doc/guides/linux_gsg/eal_args.include.rst > @@ -86,6 +86,12 @@ Multiprocessing-related options > > Set the type of the current process. > > +* ``--base-virtaddr <address>`` > + > + Attempt to use a different starting address for all memory maps of the > + primary DPDK process. This can be helpful if secondary processes cannot > + start due to conflicts in address map. > + doc/guides/freebsd_gsg/freebsd_eal_parameters.rst:.. include:: ../linux_gsg/eal_args.include.rst Ok, a bit misleading to put in linux_gsg/, so writing this here if someone else looks at this :-) > Memory-related options > ~~~~~~~~~~~~~~~~~~~~~~ > > diff --git a/doc/guides/linux_gsg/linux_eal_parameters.rst b/doc/guides/linux_gsg/linux_eal_parameters.rst > index c63f0f49a..b2cc60e44 100644 > --- a/doc/guides/linux_gsg/linux_eal_parameters.rst > +++ b/doc/guides/linux_gsg/linux_eal_parameters.rst > @@ -49,12 +49,6 @@ Multiprocessing-related options > allows running multiple independent DPDK primary/secondary processes under > different prefixes. > > -* ``--base-virtaddr <address>`` > - > - Attempt to use a different starting address for all memory maps of the > - primary DPDK process. This can be helpful if secondary processes cannot > - start due to conflicts in address map. > - > Memory-related options > ~~~~~~~~~~~~~~~~~~~~~~ > > diff --git a/doc/guides/rel_notes/release_19_08.rst b/doc/guides/rel_notes/release_19_08.rst > index 4a1fd8dd8..1b58d9282 100644 > --- a/doc/guides/rel_notes/release_19_08.rst > +++ b/doc/guides/rel_notes/release_19_08.rst > @@ -56,6 +56,11 @@ New Features > Also, make sure to start the actual text at the margin. > ========================================================= > > +* **FreeBSD now supports `--base-virtaddr` EAL option.** > + > + FreeBSD version now also supports setting base virtual address for mapping > + pages and resources into its address space. > + > * **Added MCS lock.** > > MCS lock provides scalability by spinning on a CPU/thread local variable Well, obviously, this needs some rebase on 19.11-rc0 :-) > diff --git a/lib/librte_eal/common/eal_common_options.c b/lib/librte_eal/common/eal_common_options.c > index 512d5088e..156e48e19 100644 > --- a/lib/librte_eal/common/eal_common_options.c > +++ b/lib/librte_eal/common/eal_common_options.c > @@ -20,6 +20,7 @@ > #include <rte_eal.h> > #include <rte_log.h> > #include <rte_lcore.h> > +#include <rte_memory.h> > #include <rte_tailq.h> > #include <rte_version.h> > #include <rte_devargs.h> > @@ -1095,6 +1096,36 @@ eal_parse_iova_mode(const char *name) > return 0; > } > > +static int > +eal_parse_base_virtaddr(const char *arg) > +{ > + char *end; > + uint64_t addr; > + > + errno = 0; > + addr = strtoull(arg, &end, 16); > + > + /* check for errors */ > + if ((errno != 0) || (arg[0] == '\0') || end == NULL || (*end != '\0')) > + return -1; > + > + /* make sure we don't exceed 32-bit boundary on 32-bit target */ > +#ifndef RTE_ARCH_64 > + if (addr >= UINTPTR_MAX) > + return -1; > +#endif > + > + /* align the addr on 16M boundary, 16MB is the minimum huge page > + * size on IBM Power architecture. If the addr is aligned to 16MB, > + * it can align to 2MB for x86. So this alignment can also be used > + * on x86 and other architectures. > + */ > + internal_config.base_virtaddr = > + RTE_PTR_ALIGN_CEIL((uintptr_t)addr, (size_t)RTE_PGSIZE_16M); > + > + return 0; > +} > + > /* caller is responsible for freeing the returned string */ > static char * > available_cores(void) > @@ -1408,6 +1439,13 @@ eal_parse_common_option(int opt, const char *optarg, > return -1; > } > break; > + case OPT_BASE_VIRTADDR_NUM: > + if (eal_parse_base_virtaddr(optarg) < 0) { > + RTE_LOG(ERR, EAL, "invalid parameter for --" > + OPT_BASE_VIRTADDR "\n"); > + return -1; > + } > + break; > > /* don't know what to do, leave this to caller */ > default: > diff --git a/lib/librte_eal/freebsd/eal/eal.c b/lib/librte_eal/freebsd/eal/eal.c > index d53f0fe69..7ebe11db6 100644 > --- a/lib/librte_eal/freebsd/eal/eal.c > +++ b/lib/librte_eal/freebsd/eal/eal.c > @@ -227,6 +227,14 @@ rte_eal_config_create(void) > if (internal_config.no_shconf) > return 0; > > + /* map the config before base address so that we don't waste a page */ > + if (internal_config.base_virtaddr != 0) > + rte_mem_cfg_addr = (void *) > + RTE_ALIGN_FLOOR(internal_config.base_virtaddr - > + sizeof(struct rte_mem_config), sysconf(_SC_PAGE_SIZE)); > + else > + rte_mem_cfg_addr = NULL; > + > if (mem_cfg_fd < 0){ > mem_cfg_fd = open(pathname, O_RDWR | O_CREAT, 0600); > if (mem_cfg_fd < 0) { > @@ -254,8 +262,9 @@ rte_eal_config_create(void) > return -1; > } > > - rte_mem_cfg_addr = mmap(NULL, sizeof(*rte_config.mem_config), > - PROT_READ | PROT_WRITE, MAP_SHARED, mem_cfg_fd, 0); > + rte_mem_cfg_addr = mmap(rte_mem_cfg_addr, > + sizeof(*rte_config.mem_config), PROT_READ | PROT_WRITE, > + MAP_SHARED, mem_cfg_fd, 0); > > if (rte_mem_cfg_addr == MAP_FAILED){ > RTE_LOG(ERR, EAL, "Cannot mmap memory for rte_config\n"); Nit: when compared to Linux implementation, the reattach step does not recommend using the --base-virtaddr in case the remmapping failed. > diff --git a/lib/librte_eal/linux/eal/eal.c b/lib/librte_eal/linux/eal/eal.c > index 2e5499f9b..79f5d70c3 100644 > --- a/lib/librte_eal/linux/eal/eal.c > +++ b/lib/librte_eal/linux/eal/eal.c > @@ -609,35 +609,6 @@ eal_parse_socket_arg(char *strval, volatile uint64_t *socket_arg) > return 0; > } > > -static int > -eal_parse_base_virtaddr(const char *arg) > -{ > - char *end; > - uint64_t addr; > - > - errno = 0; > - addr = strtoull(arg, &end, 16); > - > - /* check for errors */ > - if ((errno != 0) || (arg[0] == '\0') || end == NULL || (*end != '\0')) > - return -1; > - > - /* make sure we don't exceed 32-bit boundary on 32-bit target */ > -#ifndef RTE_ARCH_64 > - if (addr >= UINTPTR_MAX) > - return -1; > -#endif > - > - /* align the addr on 16M boundary, 16MB is the minimum huge page > - * size on IBM Power architecture. If the addr is aligned to 16MB, > - * it can align to 2MB for x86. So this alignment can also be used > - * on x86 */ > - internal_config.base_virtaddr = > - RTE_PTR_ALIGN_CEIL((uintptr_t)addr, (size_t)RTE_PGSIZE_16M); > - > - return 0; > -} > - > static int > eal_parse_vfio_intr(const char *mode) > { > @@ -796,16 +767,6 @@ eal_parse_args(int argc, char **argv) > internal_config.force_socket_limits = 1; > break; > > - case OPT_BASE_VIRTADDR_NUM: > - if (eal_parse_base_virtaddr(optarg) < 0) { > - RTE_LOG(ERR, EAL, "invalid parameter for --" > - OPT_BASE_VIRTADDR "\n"); > - eal_usage(prgname); > - ret = -1; > - goto out; > - } > - break; > - > case OPT_VFIO_INTR_NUM: > if (eal_parse_vfio_intr(optarg) < 0) { > RTE_LOG(ERR, EAL, "invalid parameters for --" > -- > 2.17.1 Reviewed-by: David Marchand <david.marchand@redhat.com>
On 12-Aug-19 11:19 AM, David Marchand wrote: > On Tue, Jul 16, 2019 at 1:25 PM Anatoly Burakov > <anatoly.burakov@intel.com> wrote: >> >> According to our docs, only Linuxapp supports base-virtaddr option. >> That is, strictly speaking, not true because most of the things >> that are attempting to respect base-virtaddr are in common files, >> so FreeBSD already *mostly* supports this option in practice. >> >> This commit fixes the remaining bits to explicitly support >> base-virtaddr option, and moves the arg parsing from EAL to common >> options parsing code. Documentation is also updated to reflect >> that all platforms now support base-virtaddr. >> >> Signed-off-by: Anatoly Burakov <anatoly.burakov@intel.com> >> --- >> doc/guides/linux_gsg/eal_args.include.rst | 6 +++ >> doc/guides/linux_gsg/linux_eal_parameters.rst | 6 --- >> doc/guides/rel_notes/release_19_08.rst | 5 +++ >> lib/librte_eal/common/eal_common_options.c | 38 ++++++++++++++++++ >> lib/librte_eal/freebsd/eal/eal.c | 13 ++++++- >> lib/librte_eal/linux/eal/eal.c | 39 ------------------- >> 6 files changed, 60 insertions(+), 47 deletions(-) >> >> diff --git a/doc/guides/linux_gsg/eal_args.include.rst b/doc/guides/linux_gsg/eal_args.include.rst >> index cf421a56e..ed8b0e35b 100644 >> --- a/doc/guides/linux_gsg/eal_args.include.rst >> +++ b/doc/guides/linux_gsg/eal_args.include.rst >> @@ -86,6 +86,12 @@ Multiprocessing-related options >> >> Set the type of the current process. >> >> +* ``--base-virtaddr <address>`` >> + >> + Attempt to use a different starting address for all memory maps of the >> + primary DPDK process. This can be helpful if secondary processes cannot >> + start due to conflicts in address map. >> + > > doc/guides/freebsd_gsg/freebsd_eal_parameters.rst:.. include:: > ../linux_gsg/eal_args.include.rst > > Ok, a bit misleading to put in linux_gsg/, so writing this here if > someone else looks at this :-) This was agreed upon when this file was first introduced. We don't have a "common" section and there's no real way to create it without triggering a bunch of errors in doxygen, so it was decided that putting this in a Linux GSG is the best way to do this. > > >> Memory-related options >> ~~~~~~~~~~~~~~~~~~~~~~ >> >> diff --git a/doc/guides/linux_gsg/linux_eal_parameters.rst b/doc/guides/linux_gsg/linux_eal_parameters.rst >> index c63f0f49a..b2cc60e44 100644 >> --- a/doc/guides/linux_gsg/linux_eal_parameters.rst >> +++ b/doc/guides/linux_gsg/linux_eal_parameters.rst >> @@ -49,12 +49,6 @@ Multiprocessing-related options >> allows running multiple independent DPDK primary/secondary processes under >> different prefixes. >> >> -* ``--base-virtaddr <address>`` >> - >> - Attempt to use a different starting address for all memory maps of the >> - primary DPDK process. This can be helpful if secondary processes cannot >> - start due to conflicts in address map. >> - >> Memory-related options >> ~~~~~~~~~~~~~~~~~~~~~~ >> >> diff --git a/doc/guides/rel_notes/release_19_08.rst b/doc/guides/rel_notes/release_19_08.rst >> index 4a1fd8dd8..1b58d9282 100644 >> --- a/doc/guides/rel_notes/release_19_08.rst >> +++ b/doc/guides/rel_notes/release_19_08.rst >> @@ -56,6 +56,11 @@ New Features >> Also, make sure to start the actual text at the margin. >> ========================================================= >> >> +* **FreeBSD now supports `--base-virtaddr` EAL option.** >> + >> + FreeBSD version now also supports setting base virtual address for mapping >> + pages and resources into its address space. >> + >> * **Added MCS lock.** >> >> MCS lock provides scalability by spinning on a CPU/thread local variable > > Well, obviously, this needs some rebase on 19.11-rc0 :-) Yes, will do. > >> diff --git a/lib/librte_eal/common/eal_common_options.c b/lib/librte_eal/common/eal_common_options.c >> index 512d5088e..156e48e19 100644 >> --- a/lib/librte_eal/common/eal_common_options.c >> +++ b/lib/librte_eal/common/eal_common_options.c >> @@ -20,6 +20,7 @@ >> #include <rte_eal.h> >> #include <rte_log.h> <snip> >> - rte_mem_cfg_addr = mmap(NULL, sizeof(*rte_config.mem_config), >> - PROT_READ | PROT_WRITE, MAP_SHARED, mem_cfg_fd, 0); >> + rte_mem_cfg_addr = mmap(rte_mem_cfg_addr, >> + sizeof(*rte_config.mem_config), PROT_READ | PROT_WRITE, >> + MAP_SHARED, mem_cfg_fd, 0); >> >> if (rte_mem_cfg_addr == MAP_FAILED){ >> RTE_LOG(ERR, EAL, "Cannot mmap memory for rte_config\n"); > > Nit: when compared to Linux implementation, the reattach step does not > recommend using the --base-virtaddr in case the remmapping failed. > Good catch, will fix. > >> diff --git a/lib/librte_eal/linux/eal/eal.c b/lib/librte_eal/linux/eal/eal.c >> index 2e5499f9b..79f5d70c3 100644 >> --- a/lib/librte_eal/linux/eal/eal.c >> +++ b/lib/librte_eal/linux/eal/eal.c >> @@ -609,35 +609,6 @@ eal_parse_socket_arg(char *strval, volatile uint64_t *socket_arg) >> return 0; >> } > > > Reviewed-by: David Marchand <david.marchand@redhat.com> >
Hello Anatoly, On Mon, Aug 12, 2019 at 3:38 PM Burakov, Anatoly <anatoly.burakov@intel.com> wrote: > > On 12-Aug-19 11:19 AM, David Marchand wrote: > > On Tue, Jul 16, 2019 at 1:25 PM Anatoly Burakov > > <anatoly.burakov@intel.com> wrote: > >> > >> According to our docs, only Linuxapp supports base-virtaddr option. > >> That is, strictly speaking, not true because most of the things > >> that are attempting to respect base-virtaddr are in common files, > >> so FreeBSD already *mostly* supports this option in practice. > >> > >> This commit fixes the remaining bits to explicitly support > >> base-virtaddr option, and moves the arg parsing from EAL to common > >> options parsing code. Documentation is also updated to reflect > >> that all platforms now support base-virtaddr. > >> > >> Signed-off-by: Anatoly Burakov <anatoly.burakov@intel.com> > >> --- > >> doc/guides/linux_gsg/eal_args.include.rst | 6 +++ > >> doc/guides/linux_gsg/linux_eal_parameters.rst | 6 --- > >> doc/guides/rel_notes/release_19_08.rst | 5 +++ > >> lib/librte_eal/common/eal_common_options.c | 38 ++++++++++++++++++ > >> lib/librte_eal/freebsd/eal/eal.c | 13 ++++++- > >> lib/librte_eal/linux/eal/eal.c | 39 ------------------- > >> 6 files changed, 60 insertions(+), 47 deletions(-) > >> > >> diff --git a/doc/guides/linux_gsg/eal_args.include.rst b/doc/guides/linux_gsg/eal_args.include.rst > >> index cf421a56e..ed8b0e35b 100644 > >> --- a/doc/guides/linux_gsg/eal_args.include.rst > >> +++ b/doc/guides/linux_gsg/eal_args.include.rst > >> @@ -86,6 +86,12 @@ Multiprocessing-related options > >> > >> Set the type of the current process. > >> > >> +* ``--base-virtaddr <address>`` > >> + > >> + Attempt to use a different starting address for all memory maps of the > >> + primary DPDK process. This can be helpful if secondary processes cannot > >> + start due to conflicts in address map. > >> + > > > > doc/guides/freebsd_gsg/freebsd_eal_parameters.rst:.. include:: > > ../linux_gsg/eal_args.include.rst > > > > Ok, a bit misleading to put in linux_gsg/, so writing this here if > > someone else looks at this :-) > > This was agreed upon when this file was first introduced. We don't have > a "common" section and there's no real way to create it without > triggering a bunch of errors in doxygen, so it was decided that putting > this in a Linux GSG is the best way to do this. > > > > > > >> Memory-related options > >> ~~~~~~~~~~~~~~~~~~~~~~ > >> > >> diff --git a/doc/guides/linux_gsg/linux_eal_parameters.rst b/doc/guides/linux_gsg/linux_eal_parameters.rst > >> index c63f0f49a..b2cc60e44 100644 > >> --- a/doc/guides/linux_gsg/linux_eal_parameters.rst > >> +++ b/doc/guides/linux_gsg/linux_eal_parameters.rst > >> @@ -49,12 +49,6 @@ Multiprocessing-related options > >> allows running multiple independent DPDK primary/secondary processes under > >> different prefixes. > >> > >> -* ``--base-virtaddr <address>`` > >> - > >> - Attempt to use a different starting address for all memory maps of the > >> - primary DPDK process. This can be helpful if secondary processes cannot > >> - start due to conflicts in address map. > >> - > >> Memory-related options > >> ~~~~~~~~~~~~~~~~~~~~~~ > >> > >> diff --git a/doc/guides/rel_notes/release_19_08.rst b/doc/guides/rel_notes/release_19_08.rst > >> index 4a1fd8dd8..1b58d9282 100644 > >> --- a/doc/guides/rel_notes/release_19_08.rst > >> +++ b/doc/guides/rel_notes/release_19_08.rst > >> @@ -56,6 +56,11 @@ New Features > >> Also, make sure to start the actual text at the margin. > >> ========================================================= > >> > >> +* **FreeBSD now supports `--base-virtaddr` EAL option.** > >> + > >> + FreeBSD version now also supports setting base virtual address for mapping > >> + pages and resources into its address space. > >> + > >> * **Added MCS lock.** > >> > >> MCS lock provides scalability by spinning on a CPU/thread local variable > > > > Well, obviously, this needs some rebase on 19.11-rc0 :-) > > Yes, will do. > > > > >> diff --git a/lib/librte_eal/common/eal_common_options.c b/lib/librte_eal/common/eal_common_options.c > >> index 512d5088e..156e48e19 100644 > >> --- a/lib/librte_eal/common/eal_common_options.c > >> +++ b/lib/librte_eal/common/eal_common_options.c > >> @@ -20,6 +20,7 @@ > >> #include <rte_eal.h> > >> #include <rte_log.h> > > <snip> > > >> - rte_mem_cfg_addr = mmap(NULL, sizeof(*rte_config.mem_config), > >> - PROT_READ | PROT_WRITE, MAP_SHARED, mem_cfg_fd, 0); > >> + rte_mem_cfg_addr = mmap(rte_mem_cfg_addr, > >> + sizeof(*rte_config.mem_config), PROT_READ | PROT_WRITE, > >> + MAP_SHARED, mem_cfg_fd, 0); > >> > >> if (rte_mem_cfg_addr == MAP_FAILED){ > >> RTE_LOG(ERR, EAL, "Cannot mmap memory for rte_config\n"); > > > > Nit: when compared to Linux implementation, the reattach step does not > > recommend using the --base-virtaddr in case the remmapping failed. > > > > Good catch, will fix. Can you prepare a v2? Thanks.
diff --git a/doc/guides/linux_gsg/eal_args.include.rst b/doc/guides/linux_gsg/eal_args.include.rst index cf421a56e..ed8b0e35b 100644 --- a/doc/guides/linux_gsg/eal_args.include.rst +++ b/doc/guides/linux_gsg/eal_args.include.rst @@ -86,6 +86,12 @@ Multiprocessing-related options Set the type of the current process. +* ``--base-virtaddr <address>`` + + Attempt to use a different starting address for all memory maps of the + primary DPDK process. This can be helpful if secondary processes cannot + start due to conflicts in address map. + Memory-related options ~~~~~~~~~~~~~~~~~~~~~~ diff --git a/doc/guides/linux_gsg/linux_eal_parameters.rst b/doc/guides/linux_gsg/linux_eal_parameters.rst index c63f0f49a..b2cc60e44 100644 --- a/doc/guides/linux_gsg/linux_eal_parameters.rst +++ b/doc/guides/linux_gsg/linux_eal_parameters.rst @@ -49,12 +49,6 @@ Multiprocessing-related options allows running multiple independent DPDK primary/secondary processes under different prefixes. -* ``--base-virtaddr <address>`` - - Attempt to use a different starting address for all memory maps of the - primary DPDK process. This can be helpful if secondary processes cannot - start due to conflicts in address map. - Memory-related options ~~~~~~~~~~~~~~~~~~~~~~ diff --git a/doc/guides/rel_notes/release_19_08.rst b/doc/guides/rel_notes/release_19_08.rst index 4a1fd8dd8..1b58d9282 100644 --- a/doc/guides/rel_notes/release_19_08.rst +++ b/doc/guides/rel_notes/release_19_08.rst @@ -56,6 +56,11 @@ New Features Also, make sure to start the actual text at the margin. ========================================================= +* **FreeBSD now supports `--base-virtaddr` EAL option.** + + FreeBSD version now also supports setting base virtual address for mapping + pages and resources into its address space. + * **Added MCS lock.** MCS lock provides scalability by spinning on a CPU/thread local variable diff --git a/lib/librte_eal/common/eal_common_options.c b/lib/librte_eal/common/eal_common_options.c index 512d5088e..156e48e19 100644 --- a/lib/librte_eal/common/eal_common_options.c +++ b/lib/librte_eal/common/eal_common_options.c @@ -20,6 +20,7 @@ #include <rte_eal.h> #include <rte_log.h> #include <rte_lcore.h> +#include <rte_memory.h> #include <rte_tailq.h> #include <rte_version.h> #include <rte_devargs.h> @@ -1095,6 +1096,36 @@ eal_parse_iova_mode(const char *name) return 0; } +static int +eal_parse_base_virtaddr(const char *arg) +{ + char *end; + uint64_t addr; + + errno = 0; + addr = strtoull(arg, &end, 16); + + /* check for errors */ + if ((errno != 0) || (arg[0] == '\0') || end == NULL || (*end != '\0')) + return -1; + + /* make sure we don't exceed 32-bit boundary on 32-bit target */ +#ifndef RTE_ARCH_64 + if (addr >= UINTPTR_MAX) + return -1; +#endif + + /* align the addr on 16M boundary, 16MB is the minimum huge page + * size on IBM Power architecture. If the addr is aligned to 16MB, + * it can align to 2MB for x86. So this alignment can also be used + * on x86 and other architectures. + */ + internal_config.base_virtaddr = + RTE_PTR_ALIGN_CEIL((uintptr_t)addr, (size_t)RTE_PGSIZE_16M); + + return 0; +} + /* caller is responsible for freeing the returned string */ static char * available_cores(void) @@ -1408,6 +1439,13 @@ eal_parse_common_option(int opt, const char *optarg, return -1; } break; + case OPT_BASE_VIRTADDR_NUM: + if (eal_parse_base_virtaddr(optarg) < 0) { + RTE_LOG(ERR, EAL, "invalid parameter for --" + OPT_BASE_VIRTADDR "\n"); + return -1; + } + break; /* don't know what to do, leave this to caller */ default: diff --git a/lib/librte_eal/freebsd/eal/eal.c b/lib/librte_eal/freebsd/eal/eal.c index d53f0fe69..7ebe11db6 100644 --- a/lib/librte_eal/freebsd/eal/eal.c +++ b/lib/librte_eal/freebsd/eal/eal.c @@ -227,6 +227,14 @@ rte_eal_config_create(void) if (internal_config.no_shconf) return 0; + /* map the config before base address so that we don't waste a page */ + if (internal_config.base_virtaddr != 0) + rte_mem_cfg_addr = (void *) + RTE_ALIGN_FLOOR(internal_config.base_virtaddr - + sizeof(struct rte_mem_config), sysconf(_SC_PAGE_SIZE)); + else + rte_mem_cfg_addr = NULL; + if (mem_cfg_fd < 0){ mem_cfg_fd = open(pathname, O_RDWR | O_CREAT, 0600); if (mem_cfg_fd < 0) { @@ -254,8 +262,9 @@ rte_eal_config_create(void) return -1; } - rte_mem_cfg_addr = mmap(NULL, sizeof(*rte_config.mem_config), - PROT_READ | PROT_WRITE, MAP_SHARED, mem_cfg_fd, 0); + rte_mem_cfg_addr = mmap(rte_mem_cfg_addr, + sizeof(*rte_config.mem_config), PROT_READ | PROT_WRITE, + MAP_SHARED, mem_cfg_fd, 0); if (rte_mem_cfg_addr == MAP_FAILED){ RTE_LOG(ERR, EAL, "Cannot mmap memory for rte_config\n"); diff --git a/lib/librte_eal/linux/eal/eal.c b/lib/librte_eal/linux/eal/eal.c index 2e5499f9b..79f5d70c3 100644 --- a/lib/librte_eal/linux/eal/eal.c +++ b/lib/librte_eal/linux/eal/eal.c @@ -609,35 +609,6 @@ eal_parse_socket_arg(char *strval, volatile uint64_t *socket_arg) return 0; } -static int -eal_parse_base_virtaddr(const char *arg) -{ - char *end; - uint64_t addr; - - errno = 0; - addr = strtoull(arg, &end, 16); - - /* check for errors */ - if ((errno != 0) || (arg[0] == '\0') || end == NULL || (*end != '\0')) - return -1; - - /* make sure we don't exceed 32-bit boundary on 32-bit target */ -#ifndef RTE_ARCH_64 - if (addr >= UINTPTR_MAX) - return -1; -#endif - - /* align the addr on 16M boundary, 16MB is the minimum huge page - * size on IBM Power architecture. If the addr is aligned to 16MB, - * it can align to 2MB for x86. So this alignment can also be used - * on x86 */ - internal_config.base_virtaddr = - RTE_PTR_ALIGN_CEIL((uintptr_t)addr, (size_t)RTE_PGSIZE_16M); - - return 0; -} - static int eal_parse_vfio_intr(const char *mode) { @@ -796,16 +767,6 @@ eal_parse_args(int argc, char **argv) internal_config.force_socket_limits = 1; break; - case OPT_BASE_VIRTADDR_NUM: - if (eal_parse_base_virtaddr(optarg) < 0) { - RTE_LOG(ERR, EAL, "invalid parameter for --" - OPT_BASE_VIRTADDR "\n"); - eal_usage(prgname); - ret = -1; - goto out; - } - break; - case OPT_VFIO_INTR_NUM: if (eal_parse_vfio_intr(optarg) < 0) { RTE_LOG(ERR, EAL, "invalid parameters for --"
According to our docs, only Linuxapp supports base-virtaddr option. That is, strictly speaking, not true because most of the things that are attempting to respect base-virtaddr are in common files, so FreeBSD already *mostly* supports this option in practice. This commit fixes the remaining bits to explicitly support base-virtaddr option, and moves the arg parsing from EAL to common options parsing code. Documentation is also updated to reflect that all platforms now support base-virtaddr. Signed-off-by: Anatoly Burakov <anatoly.burakov@intel.com> --- doc/guides/linux_gsg/eal_args.include.rst | 6 +++ doc/guides/linux_gsg/linux_eal_parameters.rst | 6 --- doc/guides/rel_notes/release_19_08.rst | 5 +++ lib/librte_eal/common/eal_common_options.c | 38 ++++++++++++++++++ lib/librte_eal/freebsd/eal/eal.c | 13 ++++++- lib/librte_eal/linux/eal/eal.c | 39 ------------------- 6 files changed, 60 insertions(+), 47 deletions(-)