From: Daniel Schwierzeck Date: Wed, 15 Apr 2020 14:52:53 +0200 Subject: [RFC PATCH 2/3] spl: spl_legacy: Add lzma decompression support for legacy image In-Reply-To: <20200410110431.12256-2-sr@denx.de> References: <20200410110431.12256-1-sr@denx.de> <20200410110431.12256-2-sr@denx.de> Message-ID: List-Id: MIME-Version: 1.0 Content-Type: text/plain; charset="us-ascii" Content-Transfer-Encoding: 7bit To: u-boot@lists.denx.de Am 10.04.20 um 13:04 schrieb Stefan Roese: > From: Weijie Gao > > This patch adds support for decompressing LZMA compressed u-boot payload > in legacy uImage format. > > Using this patch together with u-boot-lzma.img may be useful for some > platforms as they can reduce the size and load time of u-boot payload. > > Signed-off-by: Weijie Gao > Signed-off-by: Stefan Roese > Cc: Daniel Schwierzeck > Cc: Simon Goldschmidt > --- > common/spl/spl_legacy.c | 50 +++++++++++++++++++++++++++++++++++++++-- > 1 file changed, 48 insertions(+), 2 deletions(-) Reviewed-by: Daniel Schwierzeck nits below > > diff --git a/common/spl/spl_legacy.c b/common/spl/spl_legacy.c > index 7f00fc8885..41734c026f 100644 > --- a/common/spl/spl_legacy.c > +++ b/common/spl/spl_legacy.c > @@ -4,8 +4,15 @@ > */ > > #include > +#include > #include > > +#include > +#include > +#include > + > +#define LZMA_LEN (1 << 20) > + > int spl_parse_legacy_header(struct spl_image_info *spl_image, > const struct image_header *header) > { > @@ -55,7 +62,10 @@ int spl_parse_legacy_header(struct spl_image_info *spl_image, > int spl_load_legacy_img(struct spl_image_info *spl_image, > struct spl_load_info *load, ulong header) > { > + __maybe_unused SizeT lzma_len; > + __maybe_unused void *src; > struct image_header hdr; > + ulong dataptr; > int ret; > > /* Read header into local struct */ > @@ -65,9 +75,45 @@ int spl_load_legacy_img(struct spl_image_info *spl_image, > if (ret) > return ret; > > + dataptr = header + sizeof(hdr); > + > /* Read image */ > - load->read(load, header + sizeof(hdr), spl_image->size, > - (void *)(unsigned long)spl_image->load_addr); > + switch (image_get_comp(&hdr)) { > + case IH_COMP_NONE: > + load->read(load, dataptr, spl_image->size, > + (void *)(unsigned long)spl_image->load_addr); > + break; to avoid the little increase of binary footprint due to the extra check maybe a little wrapper like this could help: static inline int spl_image_get_comp(const struct image_header *hdr) { if (IS_ENABLED(CONFIG_SPL_LZMA) /* || IS_ENABLED(CONFIG_SPL_ANOTHER_FANCY_COMPRESSION) */) return image_get_comp(hdr); return IH_COMP_NONE; } switch (spl_image_get_comp(&hdr)) { ... } then the compiler should optimise the switch/case statement away due to Dead Code Elimination. > + > +#if IS_ENABLED(CONFIG_SPL_LZMA) > + case IH_COMP_LZMA: > + lzma_len = LZMA_LEN; > + > + debug("LZMA: Decompressing %08lx to %08lx\n", > + dataptr, spl_image->load_addr); > + src = malloc(spl_image->size); > + if (!src) { > + printf("Unable to allocate %d bytes for LZMA\n", > + spl_image->size); > + return -ENOMEM; > + } > + > + load->read(load, dataptr, spl_image->size, src); > + ret = lzmaBuffToBuffDecompress((void *)spl_image->load_addr, > + &lzma_len, src, spl_image->size); > + if (ret) { > + printf("LZMA decompression error: %d\n", ret); > + return ret; > + } > + > + spl_image->size = lzma_len; > + break; > +#endif > + > + default: > + debug("Compression method %s is not supported\n", > + genimg_get_comp_short_name(image_get_comp(&hdr))); > + return -EINVAL; > + } > > return 0; > } > -- - Daniel