ClamAV学习【4】——cli_magic_scandesc函数浏览
今晚继续浏览ClamAV代码，挖掘到了cli_magic_scandesc函数，发现前面包装了很多次扫描函数，这里就是最后一层的感觉。一些扫描限制判断加上文件类型判断，采用不同扫描函数处理。
（PS:发现这些函数个头都很大，虽然按着功能分段好理解，但是书本不是说一个函数一个功能模块么？难道是不实际，还是理解错误？）
代码分析如下：
//magic，说明这很神奇
int cli_magic_scandesc(int desc, cli_ctx *ctx)
{
int ret = CL_CLEAN;
cli_file_t type, dettype = 0;
struct stat sb;
uint8_t typercg = 1;
//获取文件属性stat结构
if(fstat(desc, &sb) == -1) {
cli_errmsg("magic_scandesc: Can't fstat descriptor %d\n", desc);
return CL_EIO;
}
//文件太小无需扫描
if(sb.st_size <= 5) {
cli_dbgmsg("Small data (%u bytes)\n", (unsigned int) sb.st_size);
return CL_CLEAN;
}
//引擎已经构建
if(!ctx->engine) {
cli_errmsg("CRITICAL: engine == NULL\n");
return CL_EMALFDB;
}
//默认模式扫描
if(!ctx->options) { /* raw mode (stdin, etc.) */
cli_dbgmsg("Raw mode: No support for special files\n");
if((ret = cli_scandesc(desc, ctx, 0, 0, NULL, AC_SCAN_VIR)) == CL_VIRUS)
cli_dbgmsg("%s found in descriptor %d\n", *ctx->virname, desc);
return ret;
}
//是否超过设置的扫描限制
if(cli_updatelimits(ctx, sb.st_size)!=CL_CLEAN)
return CL_CLEAN;
//邮件或者压缩文件扫描
if((SCAN_MAIL || SCAN_ARCHIVE) && ctx->limits && ctx->limits->maxreclevel && ctx->recursion > ctx->limits->maxreclevel) {
cli_dbgmsg("Archive recursion limit exceeded (level = %u).\n", ctx->recursion);
return CL_CLEAN;
}
//定位文件，找到文件类型
lseek(desc, 0, SEEK_SET);
type = cli_filetype2(desc, ctx->engine);
if(type == CL_TYPE_ERROR) {
cli_dbgmsg("cli_magic_scandesc: cli_filetype2 returned CL_TYPE_ERROR\n");
return CL_EIO;
}
lseek(desc, 0, SEEK_SET);
//非忽略类型还有个什么。需要研究
if(type != CL_TYPE_IGNORED && ctx->engine->sdb) {
if((ret = cli_scanraw(desc, ctx, type, 0, &dettype)) == CL_VIRUS)
return CL_VIRUS;
lseek(desc, 0, SEEK_SET);
}
ctx->recursion++;
//根据不同的类型，让文件见鬼去吧
switch(type) {
case CL_TYPE_IGNORED:
break;
case CL_TYPE_RAR:
#ifdef ENABLE_UNRAR
if(SCAN_ARCHIVE && (DCONF_ARCH & ARCH_CONF_RAR))
ret = cli_scanrar(desc, ctx, 0, NULL);
#else
cli_warnmsg("RAR code not compiled-in\n");
#endif
break;
case CL_TYPE_ZIP:
if(SCAN_ARCHIVE && (DCONF_ARCH & ARCH_CONF_ZIP))
ret = cli_unzip(desc, ctx);
break;
case CL_TYPE_GZ:
if(SCAN_ARCHIVE && (DCONF_ARCH & ARCH_CONF_GZ))
ret = cli_scangzip(desc, ctx);
break;
case CL_TYPE_BZ:
#ifdef HAVE_BZLIB_H
if(SCAN_ARCHIVE && (DCONF_ARCH & ARCH_CONF_BZ))
ret = cli_scanbzip(desc, ctx);
#endif
break;
case CL_TYPE_ARJ:
if(SCAN_ARCHIVE && (DCONF_ARCH & ARCH_CONF_ARJ))
ret = cli_scanarj(desc, ctx, 0, NULL);
break;
case CL_TYPE_NULSFT:
if(SCAN_ARCHIVE && (DCONF_ARCH & ARCH_CONF_NSIS))
ret = cli_scannulsft(desc, ctx, 0);
break;
case CL_TYPE_AUTOIT:
if(SCAN_ARCHIVE && (DCONF_ARCH & ARCH_CONF_AUTOIT))
ret = cli_scanautoit(desc, ctx, 23);
break;
case CL_TYPE_MSSZDD:
if(SCAN_ARCHIVE && (DCONF_ARCH & ARCH_CONF_SZDD))
ret = cli_scanszdd(desc, ctx);
break;
case CL_TYPE_MSCAB:
if(SCAN_ARCHIVE && (DCONF_ARCH & ARCH_CONF_CAB))
ret = cli_scanmscab(desc, ctx, 0);
break;
case CL_TYPE_HTML:
if(SCAN_HTML && (DCONF_DOC & DOC_CONF_HTML))
ret = cli_scanhtml(desc, ctx);
break;
case CL_TYPE_HTML_UTF16:
if(SCAN_HTML && (DCONF_DOC & DOC_CONF_HTML))
ret = cli_scanhtml_utf16(desc, ctx);
break;
case CL_TYPE_SCRIPT:
if((DCONF_DOC & DOC_CONF_SCRIPT) && dettype != CL_TYPE_HTML)
ret = cli_scanscript(desc, ctx);
break;
case CL_TYPE_RTF:
if(SCAN_ARCHIVE && (DCONF_DOC & DOC_CONF_RTF))
ret = cli_scanrtf(desc, ctx);
break;
case CL_TYPE_MAIL:
if(SCAN_MAIL && (DCONF_MAIL & MAIL_CONF_MBOX))
ret = cli_scanmail(desc, ctx);
break;
case CL_TYPE_TNEF:
if(SCAN_MAIL && (DCONF_MAIL & MAIL_CONF_TNEF))
ret = cli_scantnef(desc, ctx);
break;
case CL_TYPE_UUENCODED:
if(DCONF_OTHER & OTHER_CONF_UUENC)
ret = cli_scanuuencoded(desc, ctx);
break;
case CL_TYPE_MSCHM:
if(SCAN_ARCHIVE && (DCONF_ARCH & ARCH_CONF_CHM))
ret = cli_scanmschm(desc, ctx);
break;
case CL_TYPE_MSOLE2:
if(SCAN_OLE2 && (DCONF_ARCH & ARCH_CONF_OLE2))
ret = cli_scanole2(desc, ctx);
break;
case CL_TYPE_POSIX_TAR:
if(SCAN_ARCHIVE && (DCONF_ARCH & ARCH_CONF_TAR))
ret = cli_scantar(desc, ctx, 1);
break;
case CL_TYPE_OLD_TAR:
if(SCAN_ARCHIVE && (DCONF_ARCH & ARCH_CONF_TAR))
ret = cli_scantar(desc, ctx, 0);
break;
case CL_TYPE_BINHEX:
if(SCAN_ARCHIVE && (DCONF_ARCH & ARCH_CONF_BINHEX))
ret = cli_scanbinhex(desc, ctx);
break;
case CL_TYPE_SCRENC:
if(DCONF_OTHER & OTHER_CONF_SCRENC)
ret = cli_scanscrenc(desc, ctx);
break;
case CL_TYPE_RIFF:
if(SCAN_ALGO && (DCONF_OTHER & OTHER_CONF_RIFF))
ret = cli_scanriff(desc, ctx->virname);
break;
case CL_TYPE_GRAPHICS:
if(SCAN_ALGO && (DCONF_OTHER & OTHER_CONF_JPEG))
ret = cli_scanjpeg(desc, ctx->virname);
break;
case CL_TYPE_PDF: /* FIXMELIMITS: pdf should be an archive! */
if(SCAN_PDF && (DCONF_DOC & DOC_CONF_PDF))
ret = cli_scanpdf(desc, ctx, 0);
break;
case CL_TYPE_CRYPTFF:
if(DCONF_OTHER & OTHER_CONF_CRYPTFF)
ret = cli_scancryptff(desc, ctx);
break;
case CL_TYPE_ELF:
if(SCAN_ELF && ctx->dconf->elf)
ret = cli_scanelf(desc, ctx);
break;
case CL_TYPE_SIS:
if(SCAN_ARCHIVE && (DCONF_ARCH & ARCH_CONF_SIS))
ret = cli_scansis(desc, ctx);
break;
case CL_TYPE_BINARY_DATA:
ret = cli_check_mydoom_log(desc, ctx->virname);
break;
default:
break;
}
ctx->recursion--;
//没有检查植入性pe文件
if(type == CL_TYPE_ZIP && SCAN_ARCHIVE && (DCONF_ARCH & ARCH_CONF_ZIP)) {
if(sb.st_size > 1048576) {
cli_dbgmsg("cli_magic_scandesc: Not checking for embedded PEs (zip file > 1 MB)\n");
typercg = 0;
}
}
/* CL_TYPE_HTML: raw HTML files are not scanned, unless safety measure activated via DCONF */
if(type != CL_TYPE_IGNORED && (type != CL_TYPE_HTML || !(DCONF_DOC & DOC_CONF_HTML_SKIPRAW)) && ret != CL_VIRUS && !ctx->engine->sdb) {
if(cli_scanraw(desc, ctx, type, typercg, &dettype) == CL_VIRUS)
return CL_VIRUS;
}
ctx->recursion++;
lseek(desc, 0, SEEK_SET);
switch(type) {
case CL_TYPE_TEXT_ASCII:
if((DCONF_DOC & DOC_CONF_SCRIPT) && dettype != CL_TYPE_HTML)
ret = cli_scanscript(desc, ctx);
break;
/* Due to performance reasons all executables were first scanned
* in raw mode. Now we will try to unpack them
*/
case CL_TYPE_MSEXE:
if(SCAN_PE && ctx->dconf->pe)
ret = cli_scanpe(desc, ctx);
break;
default:
break;
}
ctx->recursion--;
switch(ret) {
case CL_EFORMAT:
case CL_EMAXREC:
case CL_EMAXSIZE:
case CL_EMAXFILES:
cli_dbgmsg("Descriptor[%d]: %s\n", desc, cl_strerror(ret));
return CL_CLEAN;
default:
return ret;
}
}
