Add file offset and memory alignment for the ELF segments information ("iSS") #403
Comments
|
I follow BUILDING.md and finish it. Now I use rizin to test iSS command, comparing with readelf. |
|
librz/core/cmd_info.c
…On Sun, Mar 7, 2021, 22:30 HuangPayoung ***@***.***> wrote:
I follow BUILDING.md and finish it. Now I use rizin to test iSS command,
comparing with readelf.
File offset and memory alignment are in the same struct with virtual
address, maybe I can follow the iSS code to fix it.
I read the document, but the project is big and I can't search iSS command
in document.
@XVilka <https://github.com/XVilka> @fangxlmr
<https://github.com/fangxlmr> Can you help me that which file is about
iSS command cource code?
—
You are receiving this because you were mentioned.
Reply to this email directly, view it on GitHub
<#403 (comment)>, or
unsubscribe
<https://github.com/notifications/unsubscribe-auth/AABRT7I2WOH7IL2N65H7233TCOEW5ANCNFSM4WKJBJKA>
.
|
|
Thank you. I will view the source code and learn from it. |
|
@HuangPayoung you could check also this work-in-progress pull request that does a similar thing but for the binary sections, not segments. Hope that helps. #746 |
|
I change the output as follow : [0x00407f1c]> iSS
[Segments]
nth paddr size vaddr vsize perm name offset align
――――――――――――――――――――――――――――――――――――――――――――――――――――――――――――――――――――――――――――――
0 0x00000040 0x1f8 0x00400040 0x1f8 -r-x PHDR 0x00001000 0x0
1 0x00000238 0x1c 0x00400238 0x1c -r-- INTERP 0x00001000 0x0
2 0x00000000 0x31d8c 0x00400000 0x31d8c -r-x LOAD0 0x00001000 0x0
3 0x00031e08 0x13e8 0x00631e08 0x4c90 -rw- LOAD1 0x00001000 0x0
4 0x00031e18 0x1e0 0x00631e18 0x1e0 -rw- DYNAMIC 0x00001000 0x0
5 0x00000254 0x20 0x00400254 0x20 -r-- NOTE 0x00001000 0x0
6 0x0002cc68 0xacc 0x0042cc68 0xacc -r-- GNU_EH_FRAME 0x00001000 0x0
7 0x00000000 0x0 0x00000000 0x0 -rw- GNU_STACK 0x00001000 0x0
8 0x00031e08 0x1f8 0x00631e08 0x1f8 -r-- GNU_RELRO 0x00001000 0x0
9 0x00000000 0x40 0x00400000 0x40 -rw- ehdr 0x00001000 0x0I think the next step is to add typedef struct rz_bin_section_t {
char *name;
ut64 size;
ut64 vsize;
ut64 vaddr;
ut64 paddr;
ut32 perm;
// per section platform info
const char *arch;
char *format;
int bits;
bool has_strings;
bool add; // indicates when you want to add the section to io `S` command
bool is_data;
bool is_segment;
} RzBinSection;I don't know how |
|
@HuangPayoung see the segments reading code in |
|
Today is deadline the of RSoC. |
@XVilka I think the offset field is already present, but it's just called |
|
@XVilka Since the |
|
@chinggg yes, go for it. |
|
Now I just made the similar challenges as #845 do --- a/librz/bin/p/bin_elf.inc
+++ b/librz/bin/p/bin_elf.inc
@@ -334,2 +334,3 @@ static RzList *sections(RzBinFile *bf) {
ptr->vaddr = section[i].rva;
+ ptr->align = section[i].align;
ptr->type = section[i].type;
diff --git a/librz/core/cbin.c b/librz/core/cbin.c
index a32bbfc04..7c164a219 100644
--- a/librz/core/cbin.c
+++ b/librz/core/cbin.c
@@ -2918,2 +2918,5 @@ static int bin_sections(RzCore *r, PJ *pj, int mode, ut64 laddr, int va, ut64 at
}
+ if (print_segments) {
+ rz_table_set_columnsf(table, "x", "align");
+ }
rz_table_align(table, 2, RZ_TABLE_ALIGN_RIGHT);
@@ -3028,2 +3031,5 @@ static int bin_sections(RzCore *r, PJ *pj, int mode, ut64 laddr, int va, ut64 at
}
+ if (print_segments) {
+ pj_kN(pj, "align", section->align);
+ }
pj_kN(pj, "paddr", section->paddr);
@@ -3088,2 +3094,6 @@ static int bin_sections(RzCore *r, PJ *pj, int mode, ut64 laddr, int va, ut64 at
+ if (print_segments) {
+ rz_list_append(row_list, rz_str_newf("0x%" PFMT64x, section->align));
+ }
+
rz_table_add_row_list(table, row_list);
diff --git a/librz/include/rz_bin.h b/librz/include/rz_bin.h
index 0db4af465..ff23c4bd2 100644
--- a/librz/include/rz_bin.h
+++ b/librz/include/rz_bin.h
@@ -577,2 +577,3 @@ typedef struct rz_bin_section_t {
ut32 perm;
+ ut64 align;But the align output is https://github.com/rizinorg/rizin/blob/dev/librz/bin/format/elf/elf.c#L3258-L3277 And I found the function if (!bin->shdr) {
//we don't give up search in phdr section
return get_sections_from_phdr(bin);
}@XVilka Do I understand the functions correctly? Or I need to write code manually to read or caculate the |
|
@chinggg because segments are specific to ELF and a couple other formats Rizin reuses |
|
@chinggg please also be cautious about changing how /* Here is the where all the fun starts.
* Linux kernel since 2005 calculates phdr offset wrongly
* adding it to the load address (va of the LOAD0).
* See `fs/binfmt_elf.c` file this line:
* NEW_AUX_ENT(AT_PHDR, load_addr + exec->e_phoff);
* So after the first read, we fix the address and read it again
*/
if (linux_kernel_hack && phdr_found) {
ut64 load_addr = Elf_(rz_bin_elf_get_baddr)(bin);
bin->ehdr.e_phoff = Elf_(rz_bin_elf_v2p)(bin, load_addr + bin->ehdr.e_phoff);
return read_phdr(bin, false);
} |
|
Yes, I extracted some parts of the ELF spec in #889 (review) . However, I'm not sure we should really care about the physical address, as that is not used by applications and it would hardly be a concept that can be easily abstracted in RzBinSection. |
iSS/iSSjshould contain more information about segments in case of ELF. Two most important missing parts are file offset and memory alignment:Compare that with the output of
readelf -lW:001.make.elf.x86_64.zip
The text was updated successfully, but these errors were encountered: